Search the archive.

Almost as informative as the decline in warfare has been where this decline is occurring. Traditionally, nations were constrained by opportunity. Most nations did not fight most others because they could not physically do so. Powerful nations, in contrast, tended to fight more often, and particularly to fight with other powerful states. Modern “zones of peace” are dominated by powerful, militarily capable countries. These countries could fight each other, but are not inclined to do so. At the same time, weaker developing nations that continue to exercise force in traditional ways are incapable of projecting power against the developed world, with the exception of unconventional methods, such as terrorism. The world is thus divided between those who could use force but prefer not to (at least not against each other) and those who would be willing to fight but lack the material means to fight far from home. Warfare in the modern world has thus become an activity involving weak (usually neighboring) nations, with intervention by powerful (geographically distant) states in a policing capacity. So, the riddle of peace boils down to why capable nations are not fighting each other. There are several explanations, as Mack has pointed out. The easiest, and I think the best, explanation has to do with an absence of motive. Modern states find little incentive to bicker over tangible property, since armies are expensive and the goods that can be looted are no longer of considerable value. Ironically, this is exactly the explanation that Norman Angell famously supplied before the World Wars. Yet, today the evidence is abundant that the most prosperous, capable nations prefer to buy rather than take. Decolonization, for example, divested European powers of territories that were increasingly expensive to administer and which contained tangible assets of limited value. Of comparable importance is the move to substantial consensus among powerful nations about how international affairs should be conducted. The great rivalries of the twentieth century were ideological rather than territorial. These have been substantially resolved, as Francis Fukuyama has pointed out. The fact that remaining differences are moderate, while the benefits of acting in concert are large (due to economic interdependence in particular) means that nations prefer to deliberate rather than fight. Differences remain, but for the most part the capable countries of the world have been in consensus, while the disgruntled developing world is incapable of acting on respective nations’ dissatisfaction. While this version of events explains the partial peace bestowed on the developed world, it also poses challenges in terms of the future. The rising nations of Asia in particular have not been equalbeneficiaries in the world political system. These nations have benefited from economic integration, and this has proved sufficient in the past to pacify them. The question for the future is whether the benefits of tangible resources through markets are sufficient to compensate the rising powers for their lack of influence in the policy sphere. The danger is that established powers may be slow to accommodate or give way to the demands of rising powers from Asia and elsewhere, leading to divisions over the intangible domain of policy and politics. Optimists argue that at the same time that these nations are rising in power, their domestic situations are evolving in a way that makes their interests more similar to the West. Consumerism, democracy, and a market orientation all help to draw the rising powers in as fellow travelers in an expanding zone of peace among the developed nations. Pessimists argue instead that capabilities among the rising powers are growing faster than their affinity for western values, or even that fundamental differences exist among the interests of first- and second-wave powers that cannot be bridged by the presence of market mechanisms or McDonald’s restaurants. If the peace observed among western, developed nations is to prove durable, it must be because warfare proves futile as nations transition to prosperity. Whether this will happen depends on the rate of change in interests and capabilities, a difficult thing to judge. We must hope that the optimistic view is correct, that what ended war in Europe can be exported globally. Prosperity has made war expensive, while the fruits of conflict, both in terms of tangible and intangible spoils have declined in value. These forces are not guaranteed to prevail indefinitely. Already, research on robotic warfare promises to lower the cost of conquest. If in addition, fundamental differences among capable communities arise, then warfare over ideology or policy can also be resurrected. We must all hope that the consolidating forces of prosperity prevail, that war becomes a durable anachronism.

But some researchers think this is only the beginning — that AI won’t just make a few techies wildly rich, the way social networking, smartphones, and personal computers did before. Believers in a growth explosion argue that AI is set to make society much, much richer by causing economic growth at a scale it has never experienced before. In 2020, the AI researcher Ajeya Cotra at grant maker Open Philanthropy released a report arguing that AI powerful enough to drive a surge in economic growth to 20 to 30 percent a year is coming, and more likely than not will emerge before 2100. The following year, her colleague Tom Davidson conducted a more in-depth investigation of the potential for AI to supercharge growth and concluded that per capita economic growth rates as high as 30 percent a year resulting from AI are plausible this century. This is an extremely “big if true” claim. Since good record-keeping began shortly after World War II, the US has averaged 3.2 percent economic growth per year. Since 2000, growth has been much more anemic, averaging 2.2 percent. Per capita growth — which is affected by population changes as well as economic ones — has been lower still. Nowhere before in history — not in England during the Industrial Revolution, not in Japan during its “income doubling” period in the 1960s, not in China in recent decades — has sustained growth on the scale of 20 to 30 percent per year happened. To put that number into perspective, 30 percent growth implies that the economy would double in size every 2.5 years or so. (Based on current growth levels, the US economy won’t double for 35 years.) It gets even more impressive when you take a longer view. Northwestern economist Ben Jones has noted the typical American today is about 100 times richer than humans were when economic growth began and we were all living at the edge of starvation. In a system of 30 percent growth per capita, in 25 years we’d be 1,000 times richer than we are now. Imagine everything humans have achieved since the days when we lived in caves: wheels, writing, bronze and iron smelting, pyramids and the Great Wall, ocean-traversing ships, mechanical reaping, railroads, telegraphy, electricity, photography, film, recorded music, laundry machines, television, the internet, cellphones. Now imagine accomplishing 10 times all that — in just a quarter century. This is a very, very, very strange world we’re contemplating. It’s strange enough that it’s fair to wonder whether it’s even possible. Personally, 30 percent growth is so far outside human experience to date that I have trouble even imagining what it might look like. AI could be just another useful technology, akin to a washing machine. In this view, it makes our lives a little better, like most technological improvements. But AI could also be something else entirely that would upend the assumptions we’ve used to understand the world around us for centuries. The basic case for a growth explosion In his 2021 report, Davidson lays out three general arguments for why such a dramatic explosion in economic growth might be possible. The first argument is historic. In an earlier report for Open Philanthropy, researcher David Roodman looked at the trajectory of the world economy in the very, very long run — all the way back to 10,000 BCE. He concluded that the pattern of economic growth, examined through this very wide lens, is superexponential. Exponential growth means the economy grows by a steady, compounding rate every year — 2 or 3 percent, say — like interest in your savings account. Superexponential growth means that the growth rate is increasing over time. That, Roodman concludes, is what has in fact happened. Roodman emphasizes that you should take this with several grains of salt. It’s not like we have good data on what the world economy was like in 10,000 BCE. But we do know, with a high degree of confidence, that economic growth was very slow for a very long time and then accelerated a great deal with the onset of the Industrial Revolution. That fits a superexponential story. And a superexponential story makes future increases in the rate of economic growth look very plausible. “Some people have the prior of ‘This is crazy’” when thinking about superexponential growth, Davidson told me in an interview. “And other people have the prior of ‘This has happened throughout history.’” Davidson’s second argument relies on a popular set of theories within economics for why growth has accelerated over the very long run. The short answer these theories give is that population growth enabled economic growth to speed up. “A long time ago, the world population was relatively small and the productivity of this population at producing ideas was extremely low,” Stanford economist Chad Jones explains in a 2001 paper. “Once an idea was discovered, however, consumption and fertility rose, producing a rise in population growth. More people were then available to find new ideas, and the next new idea was discovered more quickly.” Or, as Davidson summarizes: “more ideas → better farming techniques (or other innovations) → more food → more people → more ideas → ...” That feedback loop leads not just to economic growth but to accelerating economic growth. This type of theory also explains why growth has slowed down in rich countries compared to where it was in the 19th century. In a process known as the “demographic transition,” people in richer countries tend to choose, for a variety of reasons, to have fewer children. This breaks the feedback loop because more ideas leading to more food no longer necessarily leads to more people. But now, imagine that researchers are able to build two-legged robots, with hands and arms and everything, capable of performing both any physical task a human can and anything on a computer a human can. We’re talking full Blade Runner or Battlestar Galactica here (hopefully minus the rebellion). We would be able to build these robots in a much shorter time than the decades it takes to birth, raise, and educate a human worker, and at less expense. So we’d achieve much faster population growth (or at least growth in the population of working robots) and bring back the feedback loop that caused economic growth to accelerate a few centuries ago. The fast-growing population of robots would be able to come up with, and implement, enough economically useful ideas to get the economy going faster and faster and faster. The third argument for transformative growth is based on the conventional model that economists use to study growth in the medium to long run. The classic way of looking at economic growth is sometimes called the Solow-Swan model, after Robert Solow and Trevor Swan, who wrote separate papers developing it in 1956. (Solow died recently, in December 2023.) In this model, the size of the economy — the amount of goods and services being produced in a given year — depends on the amount of labor, the amount of capital, and a measure of productivity. Capital here specifically means tools and property that can be used to make stuff: machines in factories, ovens and dishwashers at restaurants, trademarks and patents that represent ideas you can use to make stuff. One of the most important aspects of this model is that there are diminishing returns to additional labor and additional capital. That’s because you need both to do anything useful. If you have a coffee shop with five baristas and no espresso machines, the first espresso machine is going to make them vastly more productive. But the 200th machine will do nothing because five baristas can’t run 200 machines simultaneously. Similarly, if you have 200 machines and no baristas, the first barista you hire is going to be enormously valuable. The 1,000th will be useless. Put human-level AI into this model and a bunch of things can happen that make superexponential growth look likely. AI could, for instance, make returns to capital constant, rather than diminishing. That’s because you can always invest in capital (namely, robots or other AI) instead of labor and get the same effect as if you’d hired someone. You can buy a robo-barista instead, and make all those espresso machines hum. That makes the labor component of growth literally irrelevant. Growth will explode. (Good.) But because demand for human labor will plunge to zero, most of humanity will be jobless and likely not share in that growth. (Bad.) Economists Philip Trammell and Anton Korinek have reviewed some 25 ways of plugging AI into this standard model, as well as more recent “endogenous” models that treat technical change differently. Many of these approaches result in a prediction of superexponential growth. Advanced AI could automate research, fueling accelerating growth in productivity. It could increase the rate of return on investment in capital by making capital more useful (you have great robots now!), which spurs people to save more, which leads to more investment in capital, and so on. The exact mechanism varies based on the model and scenario, but it’s not hard to get the models to spit out a substantial acceleration in economic growth. The models, of course, are just models, and inserting AI puts them “out of sample”: They’re designed for scenarios like the present, where human-level automation does not exist. But they’re also not just models: They express coherent stories and processes through which explosive growth could happen. It’s not hard to see how automating research, for instance, could lead to technology improving rapidly, with massive economic consequences. “There is no shortage of mechanisms through which advances in automation could have transformative growth consequences,” Trammell and Korinek conclude, “once we allow ourselves to look for them.”

Despite its enormous promise, AI is unlikely to trigger an economy-wide jump in productivity, or to support sustainable and inclusive growth, if its use is left to market forces. Achieving AI's greatest potential benefits will require a proactive two-sided approach. One is anticipating and, to the extent possible, preventing the misuse or harmful effects of the technology. The other is promoting the uses of AI that most assist and benefit people, power the economy, and help society tackle its most pressing opportunities and challenges--by making it more accessible, ensuring its widespread diffusion, and encouraging its most productivity-enhancing applications. For the moment, preventing harm and damage has received the lion's share of attention. In May, more than 350 AI industry leaders signed an open letter warning that "mitigating the risk of extinction" from AI should be a global priority alongside preventing pandemics and nuclear war; many, including one of us (Manyika), signed the letter to highlight the precautionary principle that should always be applied to powerful technology. Others have warned of the risks of misuse by bad actors with various motivations, as well as unconstrained military applications of AI in the absence of international regulations. These issues are important and should be addressed. But it is wrong to assume that simply limiting the misuse and harmful side effects of ai will ensure that its economic dividends will be delivered in a broadly inclusive way. Active policies and regulations aimed at unleashing those benefits will play a major role in determining whether AI realizes its full economic potential. First, policies will need to be developed to ensure that AI complements rather than replaces human labor. In current practice, AI tools are often developed and benchmarked against human performance, leading to an industry bias toward automation. That bias has been referred to as "the Turing trap," a term coined by Brynjolfsson, after the mathematician Alan Turing's argument that the most important test of machine intelligence is whether it can equal or surpass human performance. To get around this trap, public and private research funding for AI research should avoid an overly narrow focus on creating human-like ai. For example, in a growing number of specific tasks, AI systems can outperform humans by substantial margins, but they also require human collaborators, whose own capabilities can be further extended by the machines. More research on augmenting technologies and their uses, as well as the reorganization of workflow in many jobs, would help support innovations that use AI to enhance human productivity. Another crucial priority will be to encourage the widest possible spread of AI technologies across the economy. In the case of the earlier digital revolution, a large body of research has documented highly uneven adoption across sectors and firms. Many large employment sectors tagged, leading to a drag on productivity. This pattern could easily be repeated. In the case of generative AI, small and medium-sized firms deserve special attention, since they may not have the resources to conduct the experiments and develop use cases. It is possible that reductions in the current high costs of A I development and research, as well as competition among the major developers, will lead to affordable AI applications that can be widely implemented, by keeping costs down and spurring entrepreneurial activity. But policymakers must be diligent in creating rules that ensure that such competition results in broad diffusion and use of the technologies. A related issue is how to accelerate the use of AI by the industries that stand to benefit from them most. In many cases, some stakeholders, including employees, will understandably focus on the risks and resist adopting AI systems. To counter this tendency, policymakers and companies will need to consult with all parties involved and ensure that their interests are taken into account. At a macro level, the employment and wage effects of AI adoption--including the disappearance of some jobs even as others grow--should also be addressed. Partnerships involving government and industry and educational institutions will be needed to help people adapt to the different skill requirements needed for working in an AI-assisted environment. Income support during the transition to an AI-augmented economy may be another key ingredient, particularly in occupations such as call centers and other customer operations in which ai could put downward pressure on wages and even cause net job loss. But despite fears to the contrary, the prospect of large-scale Ai-induced unemployment does not seem likely, especially given current labor shortages in a number of sectors. Those anxieties are based on the incorrect assumption that demand is fixed, or inelastic, and hence insensitive to price and cost changes. In such a world, productivity gains automatically produce employment reductions. In fact, although there are likely to be lots of changes in the characteristics of many jobs, as well as some job displacement, overall employment levels in the economy are unlikely to change much, assuming the economy continues to grow. Research suggests that under most scenarios, more jobs will be gained than lost over the next decade or more. A larger challenge will be addressing the uneven effects of the new technologies, both within and between countries. Within countries, productivity growth is likely to be concentrated in white-collar jobs rather than blue-collar jobs because of generative Ai's particular impact on the knowledge economy. To achieve a similar productivity surge in the industrial economy, however, will require additional major advances in robotics. Despite good progress on that front, technological challenges remain, with the result that automation and augmentation in manufacturing, logistics, and autonomous vehicles are proceeding more slowly. Such a divergence in productivity growth between the knowledge economy, the wide service sector, and industrial sectors could further contribute to unequal distribution of AI gains. Countries will also need to confront the uneven adoption of advanced digital technologies both among firms within the same sector and among sectors. For example, within sectors, so-called frontier firms, which are often the most nimble, have outstripped other firms in using digital technologies. Similarly, the high-tech and financial services sectors have been faster to adopt new technologies than has health care, creating unevenness that can become a barrier to economy-wide productivity gains. Internationally, the recent breakthroughs and innovations in AI have clearly been led by the United States, with China in second place. These two countries are also home to the AI platform companies with enough computing power to train advanced LLMS. By contrast, the European Union has fallen behind the United States and China in AI, cloud computing, and other related areas. The question, then, is how quickly advanced AI applications can be implemented throughout the global economy. Under the open model that prevailed for several decades after World War II, technology could spread quite rapidly across borders. But that world is no more. The complex and increasingly restrictive constraints on flows of technology and capital--whether from the war in Ukraine, sanctions, or rising tensions between China and the United States--have created new barriers to international diffusion. Because of its digital nature, AI technology will spread; in fact, it would be very hard to stop it from doing so. But ensuring that it does so in the right way will require new forms of international economic governance. Thus, even if it lags in AI research, the EU will adopt the technology and use it. But many emerging economies will also benefit from this technology, and for them, access may be slow and uneven. The extent to which AI can be developed and used in an equitable way worldwide will determine the magnitude of its effect on the global economy. THE REAL AI CHALLENGE AI, including its most recent addition, generative AI, has the potential to produce a large and decisive upswing in productivity and growth at a moment when the global economy desperately needs it. Among many current economic challenges are supply constraints, growing pressure on overindebted countries, demographic changes, and persistent inflation, all of which threaten to limit countries' ability to sustain prosperity. With its broad scope and its ease of use, generative AI could do much to counter these forces. Moreover, the Ai revolution has unleashed an intense period of experimentation and innovation that could add much more value to the economy. But to fully realize this potential will require equally intense attention to policy. Governments, companies, and researchers will need to prioritize augmenting human skills rather than replacing them. They will need to promote the use of the technology across the whole of the economy. And they will need to build an economy in which the use of AI systems is sensitive to the needs of workers themselves and in which shocks are minimized and the widespread fears of excessive automation are addressed--or they will likely encounter unnecessary resistance. The development of AI has reached a crucial juncture. The technology's fraught potential, to bring enormous human and economic gains but also to cause very real harms, is coming sharply into focus. But harnessing the power of AI for good will require more than simply focusing on existential threats and potential damage. It will demand a positive vision of what AI can do and effective measures to turn that vision into reality. For the most likely risk that AI poses to the world today is not that it will produce some kind of civilizational catastrophe or a huge negative shock to employment. Rather, it is that without effective guidance, AI innovations could be developed and implemented in ways that simply magnify current economic disparities rather than bring about a strengthened global economy for generations to come.

Since bursting onto the public scene in 2022, generative artificial intelligence (AI) and technologies involving large language models have advanced with breathtaking speed, with implications equally fascinating and terrifying. But relative to what will unfold in 2024, what we saw in 2023 barely qualifies as a warmup. Trying to predict anything in such a fast-moving area is a mug’s game, but there are a few elements we can count on: The global generative AI market will grow robustly in the year ahead and into the early 2030s. Goldman Sachs analysts have estimated that the diffusion of AI technologies will increase global GDP by 7 percent over the next decade. 2024 will be the year when AI goes mainstream, and not just on our screens. AI systems are being embedded in the devices that define our everyday life. The result will be the arrival of the first “smartifacts”—devices with rudimentary intelligence having a greater ability to directly sense and interact with the physical world than they do now. This will begin with familiar objects—vehicles, appliances, personal electronics—but will also yield entirely new classes of devices, including dramatically more capable robots. Meanwhile, AI will also yield more prosaic surprises. Chatbots will become ubiquitous on our communications devices, displacing traditional search. Such surprises will include downsides as well. AI will play a major disruptive role in influencing public opinion and this year’s many elections around the world. Indeed, the disruptions have already begun; in September 2023, for instance, Microsoft researchers unearthed a network of Chinese-controlled social media accounts using AI to influence US voters.

The US and China are the world’s leading AI competitors, but they are also the most important AI collaborators. According to Stanford University’s 2023 Artificial Intelligence Index Report, despite increased geopolitical competition, the number of AI research collaborations between the two countries quadrupled between 2010 and 2021, though the rate of collaboration has slowed significantly since then, and will likely continue to do so. More so every day, AI is a critical domain in which the US and China cooperate, compete, and confront one another economically, technologically, politically, and militarily. The great power AI competition focuses on hardware, data, software, and talent. Each country is pushing the technology further, developing AI champions, and finding the most relevant use cases and advantageous areas for AI adoption. The US and China are, in different ways, seeking to advance their absolute and relative positions, protect their interests, and secure leverage as they each follow distinct strategies. The US is the world’s preeminent AI power, thanks to its world-leading universities and companies. Working alongside a global network of allies and partners on everything from research to export controls, Washington is concerned with keeping its advantages and accelerating the pace of domestic AI innovation. American industry leaders warn of the potential perils if the US were to “slow down American industry in such a way that China or somebody else makes faster progress.” Meanwhile, China will continue executing its long-term, state-led policies, including initiatives like “Made in China 2025” to increase self-reliance, bolster its domestic AI industry, and increase its leverage over competitors. Today, large language models (LLMs) are the primary arena of AI innovation and competition, and countries are racing to develop more advanced and capable LLMs. In general, LLM performance improves with scale – more parameters, more and better training data, more training runs and more computation. While GPT2 boasted some 1.5 billion parameters, subsequent models have grown by orders of magnitude, with GPT4 reportedly incorporating 1 trillion such variables. The massive open data ecosystem of the internet has played a fundamental role in the development of these models. The amount of compute dedicated to training these models has also increased sharply, driven by advances in silicon innovation. Finally, breakthroughs in software and model architecture design drive and accelerate improvements. While China had matched or exceeded the US’s historic lead in many fields of traditional AI research, the overwhelming majority of innovations in generative AI today are coming from the US, and China faces an uphill battle in training LLMs, for now. As with past technological revolutions, the nature of different governance systems can enhance or impede progress. Open societies like the US worry about AI risks, including the accuracy of LLM outputs and hallucinations, a phenomenon wherein an LLM provides inaccurate information based on the perception of patterns that do not exist or are otherwise faulty. But they do not let those concerns halt progress. Meanwhile, closed societies, including China, have different worries, particularly about their ability to control domestic speech and content that may not be state-approved, and their capacity to advance what many analysts describe as Beijing’s “discourse power,” a way to shape global narratives. AI systems are unpredictable, and their black-box structure features inputs that are often invisible and outputs that cannot be determined by government officials and censors. These are particular concerns for closed societies, and have led the Chinese state to impose restrictions that may limit innovation, hold back incumbents, and deter new entrants and entrepreneurs from engaging with this technology. For example, Beijing proposed new rules in April 2023 for developers and deployers of generative AI, including a requirement that “content generated using generative AI shall embody the Core Socialist Values and must not incite subversion of national sovereignty or the overturn of the socialist system.” The Party’s concern was therefore not about inaccurate information or hallucinations, but about potentially truthful outputs that may not conform to a Beijing-endorsed position. While the Chinese government has since issued somewhat less restrictive regulations, pushes for political control over technological developments highlight self-imposed challenges to AI development. As a result, Chinese LLMs are often trained on smaller, more restricted data sets than their Western counterparts. Censorship regimes may also compromise or bias access to and the production of data. And they pose even greater engineering challenges regarding moderation of non-deterministic outputs vs. democratic counterparts. In addition to models, in recent years, the AI competition between the US and China has focused on hardware, as the clearest path to greater compute is via access to high-performance graphics processing units (GPUs). While China has a robust domestic semiconductor industry, China’s vulnerability in the “chip wars,” the competition over the microchip technologies that power AI, was exposed on October 7, 2022, by a set of US-led export controls on high-end semiconductors and related technologies, which the US coordinated with the Netherlands and Japan. These export controls follow a strategy described as “small yard, high fence,” wherein a limited number of national-security sensitive technologies are protected through strict measures. While these US-led export controls have marked a clear turning point in US-China technology competition, there have been challenges to enforcement, including due to the need to coordinate with international partners over complex, global supply chains. To tighten enforcement, on October 17, 2023, the US released updated export controls on advanced semiconductor technology, specifically targeting chips and equipment essential for artificial intelligence development, including restrictions on high-performance chips and advanced lithography tools. Beijing has responded to the US-led controls with its own export controls, including on critical minerals such as germanium, gallium, and more recently graphite, areas where China has a historic supply-chain advantage. More significantly, Beijing increased its focus on domestic technological development and self-reliance. As the US-China Economic and Security Review Commission reported, Chinese companies “surged their orders for foreign semiconductor manufacturing technology in 2023, capitalizing on the roughly eight-month lag between when the Dutch government announced its intent to place controls on exports to China and its implementation.” Despite US-led export controls, China’s progress in hardware has continued, if at a slower pace. In September, the Chinese state-owned Semiconductor Manufacturing International Company (SMIC) and a Huawei subsidiary named HiSilicon demonstrated the capability to produce hardware comparable to Nvidia’s A100 chip, with the fabrication of a 7-nm chip using foreign-made equipment already present in-country. This chip is shipping in Huawei’s latest smartphone, the Mate 60 Pro, which scholar Chris Miller has described as “the most ‘Chinese’ advanced smartphone ever made” because “the phone’s primary 7-nm processor [and] many of the phone’s auxiliary chips are homegrown, including the Bluetooth, WiFi and power management chips.” The Mate 60 Pro was a significant breakthrough that showed the shortfalls of existing export controls and the capabilities of China’s semiconductor sector. While China remains two generations behind the world’s most cutting-edge chips, its technology ecosystem is a world leader in many fields, partially due to its robust domestic industries, strengths in data, and significant domestic talent pipeline. However, serious questions remain about the depth of China’s technological ecosystem, its ability to achieve comparable results without further international imports, and its capacity to scale production at acceptable costs, given the challenges that export control impose. The chip wars have significance not only for geopolitics, but also for global markets. They have reshaped the movement of goods and intellectual capital, and major chipmakers like Nvidia have attempted to operate within the new paradigm by releasing versions of their products targeted for Chinese markets that are as advanced as possible, given the restrictions. The results have been clear: US semiconductor exports to China dropped by 51%, from $6.4 billion to $3.1 billion, in the first eight months of 2023 compared to the same period in 2022. Meanwhile, the US is working to friendshore and onshore semiconductor supply chains. The CHIPS and Science Act introduced $39 billion in incentives for domestic chip production, and this funding includes national-security conditions, such as restricting recipients of federal dollars from involvement in the “material expansion” of “semiconductor manufacturing capacity” in a “foreign country of concern” (e.g., China, Russia, Iran, or North Korea). Such programs aim to give the US a relative edge and, as White House National Security Advisor Jake Sullivan stated, “as large of a lead as possible” in critical technologies. But leadership in AI does not come primarily from state-led initiatives. For countries to lead in AI, they need national strategies that foster and direct innovation, as well as world-leading AI companies and research institutions. The generative AI ecosystem will empower incumbent enterprises and also likely define the next generation of Big Tech companies. Private AI investment globally is considerable and growing, and is forecast to increase to more than $160 billion by 2025, according to Goldman Sachs Research. However, sustained growth and innovation requires a system that promotes property rights and entrepreneurship, and that provides predictable rules of the road to startups and mega-cap technology companies alike. China’s crackdown on technology companies, dating back to at least 2020, and dramatic shifts like the cancellation of a leading Chinese technology company’s IPO, increases uncertainty in the non-state sector, decreasing the dynamism of China’s technology ecosystem. The US is the preeminent AI power today. But the US-China technology competition is far from settled. We will see evolving strategies of techno-economic statecraft from both sides. In the past, China has shown a remarkable ability to overcome economic, geopolitical, and structural challenges to its position, and Beijing will continue to strive do so, as the Mate 60 Pro demonstrated. There are also potential paths of chip research and development, including advanced packaging, that provide new trajectories for improvement. For AI models, it is possible that entirely novel architectures could emerge that are not as dependent on the chips the West dominates today. Even if China remains the world’s number-two AI power, China will remain a formidable competitor to the US and to the US-led technology ecosystem. For the foreseeable future, we expect geopolitics to increasingly driving economic decision-making in Washington and Beijing, and in capitals around the world. The US, China, and other countries are using new and old tools to advance political objectives through economic means. The table below outlines a few of the most prominent techno-economic tools shaping technology competition.

In our multipolar era, G-20 meetings have taken on growing importance as a forum to air and negotiate conflicts. Witness the upcoming summit this week among the group of leading rich and developing nations in Osaka, Japan, where deep threats to the existing world order will be addressed head-on: rising tensions between the United States and Iran, a growing trade war between the world’s two largest economies, the state and strength of NATO, unpredictability in the relationship between the world’s top two nuclear powers, and a lack of rules governing increasingly borderless tech companies. At the end of June, the world’s major players could make substantial progress addressing all of them. If they do, it will fly in the face of punditry dismissing summits as increasingly pointless diplomatic exercises, evidence of a rudderless international system following the decline of Pax Americana. In the “G-Zero” world, according to this line of thinking, power has been spread too wide and thin. And too many competing participants have been invited into too many multilateral decision-making bodies, diluting their impact and importance—to the detriment of the global economy and global markets. It is certainly true that a more diffuse distribution of power reduces stability, makes anarchy more likely, and undermines the provision of global public goods—indeed, that is effectively an axiom of international relations. But that doesn’t mean that institutions like the G-20 are obsolete. In fact, the more multipolarity accelerates, the more meaningful and relevant multilateral institutions will become. This is true for two main reasons. First, given the United States’ decreased appetite for providing public goods (including trade regimes, shipping lanes, and security guarantees), other powers such as China, Japan, Russia, and even the European Union are stepping in to fill the void. And even if the body is flawed, the G-20 can provide critical coordination and conflict resolution mechanisms for such efforts. Second, and relatedly, such forums provide business and markets with critical information about how alliances and power dynamics are shifting in a more fluid international system. That’s true even if global policymaking itself is less efficient. The upcoming Osaka summit should prove an excellent example. It offers a timely opportunity to gather together every key external player in the Iran imbroglio, for example. Although formal progress toward better U.S.-Iran relations is unlikely, the summit will provide important signals on the potential for war and peace. In addition, expect host and new G-20 president Japan to coordinate anti-protectionist nations at the summit, exploiting its star role in sustaining the (now sans-United States) Trans-Pacific Partnership and its new EU-Japan trade deal to do so. At the very least, Japanese Prime Minister Shinzo Abe can provide positive atmospherics for a much-anticipated summit meeting between U.S. President Donald Trump and Chinese President Xi Jinping, the outcome of which could determine the fate, at least in the near-to-medium term, of the global economy. At the same time, Trump and Russian President Vladimir Putin are also scheduled to meet, which is relevant not only in light of the breakdown of U.S.-Russian arms treaties but also in terms of growing conflict over Russia between the White House and the U.S. Congress. Meanwhile, G-20 finance ministers appear to have made significant progress in agreeing to a new digital tax on multinational firms, a major development in the growing conflict between tech companies and the state that will likely be reinforced at the approaching summit. Recent trends in U.S. and global markets help reinforce these points. First, increasing geopolitical risk has actually accompanied a boom in U.S. and global equity markets, suggesting that declining U.S. hegemony and growing multipolarity is not necessarily bad for business—or at least for stock prices. At the same time, markets have often turned downward around the time of recent G-20 summits. At the very least, this pattern indicates the G-20 events are important to markets, suggesting that developments at G-20 meetings may wake up the market to otherwise discounted or mispriced geopolitical risks. The importance of the G-20 and bodies like it will only grow in a more multipolar world. As rare but institutionalized gatherings of major global political economies and their leaders, G-20 forums will be the place to understand shifting alliances and power dynamics. They will provide information about major geopolitical and geoeconomic uncertainties. Political and economic deals will be done and undone; conflicts revived and resolved; grievances accelerated and appeased. The result will not be order, per se. But it will certainly be more order—and more information about that order. In short: something far greater than zero.

This leads to the third danger. In the coming years, leaders will increasingly be confronted with global problems that can be managed or solved only through cooperation. As the climate crisis worsens, a Trump presidency would make a coordinated international response much harder and validate the backlash against environmental policies that has been building within advanced economies. At the same time, artificial intelligence is poised to take off, creating both valuable opportunities and enormous risks. At a moment when the United States should be turning to diplomacy to avoid wars, establish new norms, and promote greater international cooperation, the country would be led by an "America first" strongman. A TIME TO HEAL In any administration, national security policy is a peculiar mix of long-standing commitments, old political interests, new presidential initiatives, and improvised responses to sudden crises. Navigating the rough currents of the world, the Biden administration has often seemed to embody the contradictions of this dynamic, with one foot in the past, yearning nostalgically for American primacy, and one foot in the future, adjusting to the emerging world as it is. Through its affirmative agenda, the administration has reacted well to changing realities. Biden linked domestic and foreign policy through his legislative agenda. The CHIPS Act made substantial investments in science and innovation, including the domestic manufacturing of semiconductors. The act worked in parallel with ramped-up export and investment controls on China's high-tech sector, which have buttressed the United States' lead in the development of new technologies such as AI and quantum computing. Although this story is more complicated to tell than one about a tariff-based trade war, Biden's policy is in fact more coherent: revitalize U.S. innovation and advanced manufacturing, disentangle critical supply chains from China, and maintain a lead for U.S. companies in developing new and potentially transformative technologies. Biden's most significant piece of legislation, the Inflation Reduction Act, made enormous investments in clean energy technology. These investments will allow the United States to raise its ambition in meeting climate goals by pushing domestic industry and global markets to shift away from fossil fuels faster. Although this breakthrough enhanced U.S. credibility on climate change, it also created new challenges, as even allies have complained that Washington resorted to subsidies instead of pursuing coordinated cross-border approaches to reduce emissions. In this respect, however, the Biden administration was dealing with the world as it is. Congress cannot pass complex reforms such as putting a price on carbon; what it can do is pass large spending bills that invest in the United States. Despite tensions over U.S. industrial policy, the Biden administration has effectively reinvested in alliances that frayed under Trump. That effort has tacitly acknowledged that the world now features competing blocs, which makes it harder for the United States to pursue major initiatives by working through large international institutions or with other members of the great-power club. Instead, Washington has prioritized groupings of like-minded countries that are, to use a catch phrase, "fit for purpose." Collaboration with the United Kingdom and Australia on nuclear submarine technology. New infrastructure and AI initiatives through the G-7. Structured efforts to create more consultation among U.S. allies in the Indo-Pacific. This approach involves a dizzying number of parts; one can lose track of the number of regional consultative groups that now exist. But in the context of an unraveled international order, it makes sense to thread together cooperation where possible, while trying to turn new habits of cooperation into enduring arrangements. Most notably, Biden's reinvestment in European alliances paid off when Washington was able to swiftly mobilize support for Ukraine in 2022. This task was made easier by the administrations innovative release of intelligence on Russia's intentions to invade, an overdue reform of the way that Washington manages information. Although the war has reached a tenuous stalemate, the effort to fortify transatlantic institutions continues to advance. NATO has grown in size, relevance, and resourcing. European Union institutions have taken a more proactive role in foreign policy, most notably in coordinating support for Ukraine and accelerating its candidacy for EU membership. For all the understandable consternation about Washington's struggle to pass a recent aid bill for Ukraine, Europe's focus on its own institutions and capabilities was long overdue. SLOW TO CHANGE Yet there are three important ways in which the Biden administration has yet to recalibrate its approach to the world of post-American primacy. The first has to do with American politics. On several issues that engender controversy in Congress, the administration has constrained or distorted its options by preemptively deferring to outdated hard-liners. Even as Trump has demonstrated how the left-right axis has been scrambled on foreign policy, Biden at times feels trapped in the national security politics of the immediate post-9/11 era. Yet what once allowed a politician to appear tough to appease hawks in Washington was rarely good policy; now, it is no longer necessarily good politics. In Latin America, the Biden administration was slow to pivot away from Trump's "maximum pressure" campaigns on Venezuela and Cuba. Biden maintained, for example, the avalanche of sanctions that Trump imposed on Cuba, including the cynical return of that country to the State Department's list of state sponsors of terrorism just before leaving office, in January 2021. The result has been an acute humanitarian crisis in which U.S. sanctions exacerbated shortages of basic staples such as food and fuel, contributing to widespread suffering and migration. In the Middle East, the administration failed to move swiftly to reenter the politically contested Iran nuclear deal, opting instead to pursue what Biden called a "longer and stronger" agreement, even though Trump was the one who violated the deal's terms. Instead, the administration embraced Trump's Abraham Accords as central to its Middle East policy while reverting to confrontation with Iran. This effectively embraced Netanyahu's preferred course: a shift away from pursuing a two-state solution to the Israeli-Palestinian conflict and toward an open-ended proxy war with Tehran. Anyone who has worked at the nexus of US. politics and national security knows that avoiding friction with anti-Cuban and pro-Israeli hard-liners in Congress can feel like the path of least resistance. But that logic has turned into a trap. After October 7, Biden decided to pursue a strategy of fully embracing Netanyahu -- insisting (for a time) that any criticism would be issued in private and that U.S. military assistance would not be conditioned on the actions of the Israeli government. This engendered immediate goodwill in Israel, but it preemptively eliminated U.S. leverage. It also overlooked the far-right nature of Netanyahu's governing coalition, which offered warning signs about the indiscriminate way in which it planned to prosecute its military campaign, as Israeli officials cut off food and water flowing into Gaza within days of Hamas's attack. In the months that followed, the administration has been trying to catch up to a deteriorating situation, evolving from a strategy of embracing Netanyahu, to one of issuing rhetorical demands that were largely ignored, to one of partial restrictions on offensive military assistance. Ironically, by being mindful of the political risks of breaking with Netanyahu, Biden invited greater political risks from within the Democratic coalition and around the world. The temptation to succumb to Washington's outdated instincts has contributed to a second liability: the pursuit of maximalist objectives. The administration has shown some prudence in this area. Even as competition ramped up with China, Biden has worked over the last year to rebuild lines of communication with Beijing and has largely avoided provocative pronouncements on Taiwan. And even as he committed the United States to helping Ukraine defend itself, Biden set the objective of avoiding a direct war between the United States and Russia (although his rhetoric did drift into endorsing regime change in Moscow). The bigger challenge has at times come from outside the administration, as some supporters of Ukraine indulged in a premature triumphalism that raised impossible expectations for last year's Ukrainian counteroffensive. Paradoxically, this impulse ended up hurting Ukraine: when the campaign inevitably came up short, it made the broader U.S. policy toward Ukraine look like a failure. Sustaining support for Ukraine will require greater transparency about what is achievable in the near term and an openness to negotiations in the medium term. Gaza also showcases the danger of maximalist aims. Israel's stated objective of destroying Hamas has never been achievable. Since Hamas would never announce its own surrender, pursuing this goal would require a perpetual Israeli occupation of Gaza or the mass displacement of its people. That outcome may be what some Israeli officials really want, as evidenced by right-wing ministers' own statements. It is certainly what many people around the world, horrified by the campaign in Gaza, believe the Israeli government really wants. These critics wonder why Washington would support such a campaign, even as its own rhetoric opposes it. Instead of seeking to moderate Israel's unsustainable course, Washington needs to use its leverage to press for negotiated agreements, Palestinian state building, and a conception of Israeli security that is not beholden to expansionism or permanent occupation. Indeed, too many prescriptions sound good in Washington but fail to account for simple realities. Even with the United States' military advantage, China will develop advanced technologies and maintain its claim over Taiwan. Even with sustained U.S. support, Ukraine will have to live next to a large, nationalist, nuclear-armed Russia. Even with its military dominance, Israel cannot eliminate the Palestinian demand for self-determination. If Washington allows foreign policy to be driven by zero-sum maximalist demands, it risks a choice between open-ended conflict and embarrassment. This leads to the third way in which Washington must change its approach. Too often, the United States has appeared unable or unwilling to see itself through the eyes of most of the world's population, particularly people in the global South who feel that the international order is not designed for their benefit. The Biden administration has made laudable efforts to change this perception -- for instance, delivering c oviD-19 vaccines across the developing world, mediating conflicts from Ethiopia to Sudan, and sending food aid to places hit hard by shortages exacerbated by the war in Ukraine. Yet the overuse of sanctions, along with the prioritization of Ukraine and other U.S. geopolitical interests, misreads the room. To build better ties with developing countries, Washington needs to consistently prioritize the issues they care about: investment, technology, and clean energy. Once again, Gaza interacts with this challenge. To be blunt: for much of the world, it appears that Washington doesn't value the lives of Palestinian children as much as it values the lives of Israelis or Ukrainians. Unconditional military aid to Israel, questioning the Palestinian death toll, vetoing cease-fire resolutions at the UN Security Council, and criticizing investigations into alleged Israeli war crimes may all feel like autopilot in Washington -- but that's precisely the problem. Much of the world now hears U.S. rhetoric about human rights and the rule of law as cynical rather than aspirational, particularly when it fails to wrestle with double standards. Total consistency is unattainable in foreign policy. But by listening and responding to more diverse voices from around the world, Washington could begin to build a reservoir of goodwill. A FAREWELL TO PRIMACY In its more affirmative agenda, the Biden administration is repositioning the United States for a changing world by focusing on the resilience of its own democracy and economy while rebooting alliances in Europe and Asia. To extend that regeneration into something more global and lasting, it should abandon the pursuit of primacy while embracing an agenda that can resonate with more of the world's governments and people. As was the case in the Cold War, the most important foreign policy achievement will simply be avoiding World War III. Washington must recognize that all three fault lines of global conflict today -- Russia-Ukraine, Iran-Israel, and China-Taiwan -- run across territories just beyond the reach of U.S. treaty obligations. In other words, these are not areas where the American people have been prepared to go to war directly. With little public support and no legal obligation to do that, Washington should not count on bluffing or military buildups alone to resolve these issues; instead, it will have to focus relentlessly on diplomacy, buttressed by reassurance to frontline partners that there are alternative pathways to achieving security. In Ukraine, the United States and Europe should focus on protecting and investing in the territory controlled by the Ukrainian government -- drawing Ukraine into European institutions, sustaining its economy, and fortifying it for lengthy negotiations with Moscow so that time works in Kyiv's favor. In the Middle East, Washington should join with Arab and European partners to work directly with Palestinians on the development of new leadership and toward the recognition of a Palestinian state, while supporting Israel's security. Regional de-escalation with Iran should, as it did during the Obama administration, begin with negotiated restrictions on its nuclear program. In Taiwan, the United States should try to preserve the status quo by investing in Taiwanese military capabilities while avoiding saber rattling, by structuring engagement with Beijing to avoid miscalculation, and by mobilizing international support for a negotiated, peaceful resolution to Taiwan's status. Hawks will inevitably attack diplomacy on each of these issues with tired charges of appeasement, but consider the alternative of seeking the total defeat of Russia, regime change in Iran, and Taiwanese independence. Can Washington, or the world, risk a drift into global conflagration? Moreover, the reality is that sanctions and military aid alone will not stop war from spreading or somehow cause the governments of Russia, Iran, and China to collapse. Better outcomes, including within those countries, will be more attainable if Washington takes a longer view. Ultimately, the health of the United States' own political model and society is a more powerful force for change than purely punitive measures. Indeed, one lesson that is lost on today's hawks is that the civil rights movement did far more to win the Cold War than the war in Vietnam did. None of this will be easy, and success is not preordained, since unreliable adversaries also have agency. But given the stakes, it is worth exploring how a world of competing superpower blocs could be knitted into coexistence and negotiation on issues that cannot be dealt with in isolation. For instance, AI presents one area in which nascent dialogue between Washington and Beijing should evolve into the pursuit of shared international norms. Laudable U.S. efforts to pursue collaborative research on AI safety with like-minded countries will inevitably have to expand to further include China in higher-level and more consequential talks. These efforts should seek agreement on the mitigation of extreme harms, from the use of Ai in developing nuclear and biological weapons to the arrival of artificial general intelligence, an advanced form of AI that risks surpassing human capacities and controls. At the same time, as AI moves out into the world, the United States can use its leadership to work with countries that are eager to harness the technology for positive ends, particularly in the developing world. The United States could offer incentives for countries to cooperate with Washington on both AI safety and affirmative uses of new technologies.

Beijing, of course, has no intention of ceding technological dominance to Washington. It is working hard to develop its own advanced AI military applications. China is investing heavily in many of the same AI use cases as the United States--such as surveillance, target identification, and drone swarms. The difference is that it may not be bound by the same ethical constraints as the United States and its allies, particularly when it comes to using fully autonomous weapons systems. In the race for technological supremacy, China has some obvious advantages. Unlike Washington, Beijing can dictate its country's economic priorities and allocate whatever resources it deems necessary to meet Ai targets. China's national security policy encourages Chinese hackers, officials, and employees to steal Western intellectual property, and Beijing is unabashed in trying to recruit leading Western technologists to work with Chinese institutions. Because China has a policy of "civil-military fusion," which eliminates barriers between its civilian and military sectors, the People's Liberation Army can draw on the work of Chinese experts and companies whenever it likes. And by 2025, China will churn out nearly twice as many Ph.D. candidates in science, technology, engineering, and mathematics as the United States does, flooding China's economy with talented computer scientists in particular. But the United States has its own unique strengths. The country's market-based economy and more open political system give developers room to be creative. It has unrivaled innovation ecosystems in Silicon Valley, the Austin metropolitan area, the Massachusetts Route 128 corridor, and elsewhere. The United States also has a vibrant venture capital and private equity ecosystem that draws incomparable domestic and international investment. It is home to many of the world's leading universities, allowing it to attract and retain some of the world's best tech talent. Indeed, half the startups in Silicon Valley have at least one founder who is an immigrant. Even among those who lament China's rapid AI progress, few, if any, would trade the United States' hand for China's. But almost all of them would agree the United States needs to play its hand better to win. To do so, the Defense Department and the intelligence community will have to invest more in accelerating AI adoption. They can start by building common digital infrastructure systems that share the same standards to ensure interoperability. The infrastructure would include cloud-based technologies and services; common data standards; validated data sets; shared access to secure software stacks; sophisticated tools for the testing, evaluation, and validation of AI models; and secure application programming interfaces that control who gets access to what information at various levels of classification. The goal would be to give developers the data, algorithms, tools, and compute power--or high-speed computing power--they need to create, test, validate, and use new AI tools. Those tools will only be as good as the people who operate them, of course, and right now, the Defense Department does not have a digitally adept workforce. Few people on staff understand enough about Ai to properly govern its use, to test and evaluate AI tools to ensure they meet the Pentagon's "responsible AI" standards, or to assess which a I models best meet the needs of the military or the Defense Department--one of the world's largest enterprises. To attract more AI talent and to make better use of the tech workforce it already has, the Defense Department will need to improve how it recruits and manages digitally skilled employees. The Pentagon can start by following the advice of the National Security Commission on AI and establish a digital corps (modeled on the Army Medical Corps) to organize, train, and equip technologists. In addition, all the existing military service academies should start teaching the basics of AI, and the Pentagon should also establish a U.S. digital service academy that would educate and train aspiring civilian technologists, offering them a free college education in exchange for a commitment to serve in government for at least five years after graduating. Finally, the Defense Department should create a digital reserve corps in which tech workers from across the United States could volunteer, part time, to serve their country. The Pentagon, however, will never be able to attract as many AI experts as the private sector. The defense establishment must therefore improve how it leverages outside talent. For starters, the Defense Department should deepen its conversations with technology companies and the computer science departments of leading universities. It should also reduce some of the outdated barriers to tech firms doing business with the government. To do so, defense officials must rethink how they buy software-based products and services, including AI . Instead of taking years to develop a fixed set of highly specific requirements--as the department does when procuring military hardware--it should quickly identify the specific problems it is trying to solve and the common standards that any proposed solutions must meet and then allow companies to offer solutions in a competitive bidding process. It should also make sure that the people who will actually use the specific AI tools are able to provide feedback as models are being developed and tested. In fact, the Pentagon should create a dedicated career path for acquisition professionals who want to specialize in AI and other commercially driven technologies. Most of the Defense Department's current acquisition corps have been trained to buy complex weapons systems, such as submarines, missiles, and jets, which requires paying meticulous attention to whether contractors meet rigid specifications, cost requirements, and scheduled milestones. As a result, most of these professionals are (understandably) highly risk averse--they are neither trained nor incentivized to buy rapidly developing commercial technologies or to disrupt an existing multiyear acquisition program to integrate a more effective new technology. The Pentagon should therefore create a new cohort of acquisition experts who are specifically trained to buy these kinds of systems. This cadre should be considered the Green Berets of the acquisition force, and its members should be rewarded and promoted based on their ability to quickly deliver and scale needed commercial technologies, such as Ai. Although internal reforms will help the Pentagon accelerate progress, defense officials will also need sustained congressional support to keep pace with their Chinese counterparts. To that end, Congress should give the Defense Department more flexible funding that allows it to optimally manage AI programs. Most of the Pentagon's appropriations are fixed: when Congress funds a program, the department cannot simply redirect the money to something else. But Ai is evolving so fast, and in so many different directions, that defense officials need more reprogramming authorities and more flexible funding so they can quickly move money out of under performing projects and reinvest it in more promising ones, giving Congress appropriate notice. This approach is critical to enabling the Pentagon to adopt AI with more agility and speed. Congress should simultaneously provide the Chief Digital and Artificial Intelligence Office with bridge funding to help promising AI pilot projects cross the so-called valley of death--the difficult period between when a project demonstrates success and when the department is ready to make it a full-scale program of record. The U.S. military simply cannot afford to delay the adoption of a critical AI tool that emerges in 2023 until the 2025 budget or later. The United States will also need to continue attracting the best tech talent in the world, including by reforming elements of the U.S. immigration system. Science and technology students and workers may want to come to and stay in the United States, but byzantine immigration rules make it impossible for many of them to do so. Educational visas, for instance, do not let foreign students stay in the United States for more than three years after graduation. The resulting dynamic is perverse: U.S. institutions train many of the world's best tech experts, only to send them away. Many of them are Chinese and return to China. In addition, congressionally imposed caps on H-1B visas--the visa the United States most commonly offers to skilled workers--mean that the country can bring in only a small percentage of people who apply. For example, from the 758,994 eligible electronic registrations received during the 2023 H-1B lottery, only 110,791 people were selected (or less than 15 percent). In short, the United States is keeping out much-needed foreign talent that would willingly and meaningfully contribute to the country's ability to compete in Ai and other critical technologies. HIGH RISK, HIGH REWARD AI is indispensable to the United States' future security. But it also poses major risks. Ai is already accelerating the spread of disinformation online and facilitating inadvertent discrimination in hiring. Computer scientists have argued that it could enable automated cyberattacks at "machine speeds," as well. Chemists have shown that AI can synthesize chemical weapons, and biologists have expressed concern that it could be used to design new pathogens or bioweapons. The risks are severe enough that even AI industry leaders have expressed alarm. In May 2023, the heads of almost every major U.S. AI lab signed a letter warning that their inventions could pose an existential threat to humanity. Indeed, national security is the realm of human activity where the risks of AI are most profound. Ai models could, for example, misidentify people or objects as targets, resulting in unintended death and destruction during conflict. Black box AI models--ones whose reasoning cannot be adequately understood or explained--might lead military planners to make hazardous decisions. This risk would be most acute if AI developed for one situation were applied to another without enough testing and oversight. What might be perfectly rational and responsible in one situation might be irrational and dangerous in another. The risks do not stem just from poorly designed or carelessly used systems. The United States could be fastidious in developing and implementing AI, only for its adversaries to find ways to corrupt U.S. data, prompting systems to go haywire. For example, if an adversary were able to spoof an AI-enabled computer vision tool into targeting a civilian vehicle instead of a military one, it could cause the United States to inadvertently harm civilians in a conflict zone, undermining U.S. credibility and moral authority. An adversary could also corrupt data in ways that would degrade the performance of an AI-enabled weapon system or that could cause it to shut down. The Pentagon is aware of these risks, and in February 2020, it issued a set of ethical principles governing how AI should be used. One principle called on the department's personnel to exercise judgment and care in developing, deploying, and using AI capabilities. Another said the Defense Department will try to "minimize unintended bias in AI capabilities." A third called for ensuring that all AI is made and used in ways that can be understood and explained--with data and methodologies that are transparent and auditable. And defense leaders have directed their employees to make sure that AI systems are rigorously tested for their safety, security, and effectiveness; that AI systems are assigned to clearly defined uses; and that AI systems can be disengaged or deactivated if they exhibit unintended behavior. For autonomous and semiautonomous weapons, the Defense Department has issued even more specific guidance. Pentagon leaders have directed commanders and operators to use careful judgment over AI-enabled weapons, including by ensuring that these weapons are used in ways that are consistent with the parameters of the model's training and with the rules of engagement for the operation in which the AI is being deployed. The Defense Department's rules also stipulate that commanders use AI in accordance with the laws of war. For example, any AI-enabled weapon must be discriminate, able to distinguish between combatants and noncombatants on the battlefield, and able to avoid deliberately targeting the latter. The Pentagon has also forbidden the use of AI in its nuclear command-and-control systems, and it has urged other nuclear powers to do the same. Among the U.S. defense community's leadership, these "responsible AI" rules have achieved great consensus. But putting them into practice is no small challenge--especially given the size of the United States' defense apparatus. The Pentagon has started the process by creating a high-level governance body, beginning to establish data and digital infrastructure to support a variety of AI applications; building out the testing, evaluation, and validation capabilities needed to ensure compliance with the Defense Department's AI principles; and increasing AI awareness across the department. This implementation process is still in its infancy. But the policy framework provides a sound basis on which to build. Still, the Pentagon would be wise to further strengthen these guidelines. For example, defense officials should require that all AI vendors give the Defense Department full transparency into the origins of data they use in their training sets. In addition, the department should make sure that the behavior of any AI model it adopts is explainable (fully understood by its users and developers), without stifling innovation. It can do so by strengthening how it tests, evaluates, and verifies systems. The department should also scale and broaden the work done by the Defense Advanced Research Projects Agency--one of the entities responsible for developing emerging technologies for the military--on making sure that AI tools are explainable and responsible by design. The department's ethical principles, in other words, should be treated as required traits that shape how defense AI models are designed from the start. But the US. defense community will not be able to speed AI adoption unless the public believes it Will Use AI in ways that are effective, responsible, ethical, and lawful. Otherwise, the first time an Ai application leads to a very bad decision or serious unintended consequences on the battlefield, warfighters are unlikely to trust it, and policymakers and lawmakers are likely to suspend or prohibit its use. The Defense Department must therefore increase its investment in the research and development of AI safety and security. It must be transparent about what it will and will not use AI to do. And the Pentagon should consider making its vendors put guardrails on how they develop AI. If a company wants to provide AI to the military, for example, the Defense Department could require it to meet rigorous data protection and cybersecurity standards. By doing so, the Pentagon could help make ai safer, not just for the armed forces, but for everyone. The United States, of course, cannot single handedly make sure that ai is developed and used responsibly. Other countries--including competitors--will also have to adopt policy guardrails and norms. The world took a valuable first step when, in November 2021,193 countries approved a global agreement on the ethics of artificial intelligence--the world's first. It includes the principle that countries must guarantee human oversight of and agency over all ai. Although this agreement is an important foundation, the United States should seek out venues to discuss AI with its potential adversaries, especially China, just as it found ways to talk about nuclear weapons and other forms of arms control with the Soviet Union during the Cold War. To succeed, Washington will also have to work closely with its allies and partners to make sure they are all on the same page. Countries that agree on a set of AI norms should be willing to threaten violators with severe costs, including multilateral economic sanctions, expulsion from international forums, and legal action to hold perpetrators responsible for damage. Actors that violate Ai rules, for instance, could be indicted in a U. S. federal court, as five Chinese hackers were in 2014 for launching cyberattacks on U.S. companies. States that violate these rules could face potential retaliation for any harm done--including, in extreme cases, military action. THE NEED FOR RESPONSIBLE SPEED In the world of microelectronics, experts often talk about Moore's law: the principle that the number of transistors on chips doubles every two years, resulting in exponentially more capable devices. The law helps explain the rapid rise of so many technological innovations, including smartphones and search engines. Within national security, AI progress has created another kind of Moore's law. Whichever military first masters organizing, incorporating, and institutionalizing the use of data and AI into its operations in the coming years will reap exponential advances, giving it remarkable advantages over its foes. The first adopter of AI at scale is likely to have a faster decision cycle and better information on which to base decisions. Its networks are likely to be more resilient when under attack, preserving its ability to maintain situational awareness, defend its forces, engage targets effectively, and protect the integrity of its command, control, and communications. It will also be able to control swarms of unmanned systems in the air, on the water, and under the sea to confuse and overwhelm an adversary. The United States cannot afford to fall behind. But the national security apparatus cannot afford to be reckless, either. Without proper safeguards, AI models could cause all kinds of unintended harm. Rogue systems could even kill U.S. troops or unarmed civilians in or near areas of combat. The United States therefore finds itself facing a conundrum. The stakes of slowing AI down are unacceptably high, but so are the stakes of racing ahead without needed precautions. U.S. policymakers appear to understand this paradox. Congressional leaders know that if they were to regulate AI with too heavy a hand, they could prompt the best AI innovators to leave the United States to work where there are fewer restrictions, and the United States would then fall behind its competitors. But both Democratic and Republican policymakers also know that some regulation and oversight is essential to ensuring that AI adoption is safe and responsible. The House of Representatives and the Senate are holding sessions to educate their members and scheduling hearings to get advice from experts. These efforts to build bipartisan consensus before legislating should be applauded. Yet understanding the problem is just the first step. To solve it--to balance the need for speed with the need for safety--policymakers will have to implement better approaches to accelerating adoption as well as ensuring safety. Otherwise, Americans risk being caught in a world of both spiraling AI dangers and declining U.S. power and influence.

Escalating competition between the US and China, wars in Europe and the Middle East, and shifting global alliances have ushered in the most unstable geopolitical period since the Cold War. At the same time, we are experiencing what may be the most significant innovation since the internet: the rise of generative artificial intelligence. With the public release of ChatGPT on November 30, 2022, the defining geopolitical and technological revolutions of our time collided. Over the last 20 years, “traditional AI” has become pervasive, powering advertising algorithms, content recommendations, and social-media targeting. But its power and influence were largely shrouded and embedded within applications. Now, generative AI is front-and-center, with user interfaces that are commonplace, proficient, and clearly identifiable as machine-driven intelligence. Unlike previous technological revolutions, from the printing press to the internet, leaders in every capital and commercial center gained access to this tool at the same time, as ChatGPT became the fastest-adopted technology in history. While the technology’s future is uncertain, generative AI was nearly universally acknowledged as a paradigm-shifting innovation, not a fleeting trend or hype cycle. With widespread adoption and accelerating innovation, we have now entered a period we call the inter-AI years, when leaders in every sector are working to understand what generative AI will mean for them, and how they can take advantage of opportunities while mitigating risks. This simultaneity does not imply parity. Different countries and companies have diverse histories, contexts, capabilities, and risk tolerance when it comes to AI. ­Notwithstanding these differences, leaders have the same window of maximum flexibility to shape an AI-enabled future. This window will be brief – a few years at most – then views and strategies will harden, and norms, values, and standards will be embedded within the technology. While the technology will continue to progress, decisions made today will determine what is possible in the future. Countries will have devised their approaches and formed likeminded blocs, competition will increase, the costs of changing course will rise, and the available paths for companies and countries will be more determined. What will emerge is a generative world order. We are at the beginning of that order, and its future is uncertain. The long-term effects of technological revolutions are not made clear overnight – the Reformation did not immediately follow the invention of the printing press, and few could have predicted the positive and negative effects of social media in the early 2000s. But before long, the way we live and the nature of global politics will be shaped by generative AI. Which winners and losers will emerge? Will generative AI prove to be a zero-sum game, or create win-win scenarios? Will the problems that generative AI creates be solved by the very same technology, or will they prove to be intractable? Will companies or governments drive the future of AI? The relative nascency of this new node of AI innovation and the steep trajectory of improvement in generative AI models make these questions more momentous, and difficult to answer. We believe that this is the time to take stock of where we are, and to identify some of the inflection points that will shape our AI future. This paper takes up the key changes in the generative AI field over the last year in the domains of geopolitics, technology, and markets. We discuss what trends are emerging, what debates remain unsettled, and how the generative world order is being defined. AI and geopolitics Geopolitical competition is a constant. But the technologies that animate that competition are not. And while the United States, China, and Russia do not agree on many things, they all acknowledge that AI could reshape the balance of power. Today’s geopolitical rivals are putting AI at the center of their national strategies. In 2017, Russian President Vladimir Putin said, “Artificial intelligence is the future not only of Russia but of all of mankind.” Five years later, Chinese Communist Party General Secretary Xi Jinping declared, “We will focus on national strategic needs, gather strength to carry out indigenous and leading scientific and technological research, and resolutely win the battle in key core technologies.” And, in 2023, US President Joe Biden summarized what AI will mean for humanity: “We’re going to see more technological change in the next 10 – maybe the next 5 years – than we’ve seen in the last 50 years…Artificial intelligence is accelerating that change.” Today, there are two AI superpowers, the United States and China. But they are not the only countries that will define the technology’s future. Earlier this year, we identified a category of geopolitical swing states – non-great powers with the capacity, agency, and increasingly the will to assert themselves on the global stage. Many of these states have the power to meaningfully shape the future of AI. There are also emerging economies that have the potential to reap the rewards of AI – if the right policies and institutions are established – and whose talent, resources, and voices are essential to ensure that the creation of a human-like intelligence benefits all of humanity.

But with the latest types of artificial intelligence, the computer program is no longer a tool; it actually makes many of the decisions involved in the creative process without human intervention. COMMERCIAL IMPACT One could argue that this distinction is not important, but the manner in which the law tackles new types of machine-driven creativity could have far-reaching commercial implications. Artificial intelligence is already being used to generate works in music, journalism and gaming. These works could in theory be deemed free of copyright because they are not created by a human author. As such, they could be freely used and reused by anyone. That would be very bad news for the companies selling the works. Imagine you invest millions in a system that generates music for video games, only to find that the music is not protected by law and can be used without payment by anyone in the world. While it is difficult to ascertain the precise impact this would have on the creative economy, it may well have a chilling effect on investment in automated systems. If developers doubt whether creations generated through machine learning qualify for copyright protection, what is the incentive to invest in such systems? On the other hand, deploying artificial intelligence to handle time-consuming endeavors could still be justified, given the savings accrued in personnel costs, but it is too early to tell. LEGAL OPTIONS There are two ways in which copyright law can deal with works where human interaction is minimal or non-existent. It can either deny copyright protection for works that have been generated by a computer or it can attribute authorship of such works to the creator of the program. To my knowledge, conferring copyright in works generated by artificial intelligence has never been specifically prohibited. However, there are indications that the laws of many countries are not amenable to non-human copyright. In the United States, for example, the Copyright Office has declared that it will “register an original work of authorship, provided that the work was created by a human being.” This stance flows from case law (e.g. Feist Publications v Rural Telephone Service Company, Inc. 499 U.S. 340 (1991)) which specifies that copyright law only protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Similarly, in a recent Australian case (Acohs Pty Ltd v Ucorp Pty Ltd), a court declared that a work generated with the intervention of a computer could not be protected by copyright because it was not produced by a human. In Europe the Court of Justice of the European Union (CJEU) has also declared on various occasions, particularly in its landmark Infopaq decision (C-5/08 Infopaq International A/S v. Danske Dagbaldes Forening), that copyright only applies to original works, and that originality must reflect the “author’s own intellectual creation.” This is usually understood as meaning that an original work must reflect the author’s personality, which clearly means that a human author is necessary for a copyright work to exist. The second option, that of giving authorship to the programmer, is evident in a few countries such as the Hong Kong (SAR), India, Ireland, New Zealand and the UK. This approach is best encapsulated in UK copyright law, section 9(3) of the Copyright, Designs and Patents Act (CDPA), which states: “In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.” Furthermore, section 178 of the CDPA defines a computer-generated work as one that “is generated by computer in circumstances such that there is no human author of the work”. The idea behind such a provision is to create an exception to all human authorship requirements by recognizing the work that goes into creating a program capable of generating works, even if the creative spark is undertaken by the machine. ADDRESSING AMBIGUITY This leaves open the question of who the law would consider to be the person making the arrangements for the work to be generated. Should the law recognize the contribution of the programmer or the user of that program? In the analogue world, this is like asking whether copyright should be conferred on the maker of a pen or the writer. Why, then, could the existing ambiguity prove problematic in the digital world? Take the case of Microsoft Word. Microsoft developed the Word computer program but clearly does not own every piece of work produced using that software. The copyright lies with the user, i.e. the author who used the program to create his or her work. But when it comes to artificial intelligence algorithms that are capable of generating a work, the user’s contribution to the creative process may simply be to press a button so the machine can do its thing. There are already several text-generating machine learning programs out there, and while this is an ongoing area of research, the results can be astounding. Stanford PhD student Andrej Karpathy taught a neural network how to read text and compose sentences in the same style, and it came up with Wikipedia articles and lines of dialogue that resembled the language of Shakespeare. Some case law seems to indicate that this question could be solved on a case-by-case basis. In the English case of Nova Productions v. Mazooma Games [2007] EWCA Civ 219, the Court of Appeal had to decide on the authorship of a computer game, and declared that a player’s input “is not artistic in nature and he has contributed no skill or labour of an artistic kind”. So considering user action case by case could be one possible solution to the problem. THE FUTURE Things are likely to become yet more complex as use of artificial intelligence by artists becomes more widespread, and as the machines get better at producing creative works, further blurring the distinction between artwork that is made by a human and that made by a computer. Monumental advances in computing and the sheer amount of available computational power may well make the distinction moot; when you give a machine the capacity to learn styles from large datasets of content, it will become ever better at mimicking humans. And given enough computing power, soon we may not be able to distinguish between human-generated and machine-generated content. We are not yet at that stage, but if and when we do get there, we will have to decide what type of protection, if any, we should give to emergent works created by intelligent algorithms with little or no human intervention. Although copyright laws have been moving away from originality standards that reward skill, labor and effort, perhaps we can establish an exception to that trend when it comes to the fruits of sophisticated artificial intelligence. The alternative seems contrary to the justifications for protecting creative works in the first place. Granting copyright to the person who made the operation of artificial intelligence possible seems to be the most sensible approach, with the UK’s model looking the most efficient. Such an approach will ensure that companies keep investing in the technology, safe in the knowledge that they will get a return on their investment. The next big debate will be whether computers should be given the status and rights of people, but that is a whole other story.

Ensuring that copyright doesn’t protect ideas is not just embedded in the case law and the Copyright Act. It is a fundamental part of what makes copyright law constitutional. Along with the fair use defense, the Supreme Court tells us the idea- expression dichotomy prevents copyright from treading on the First Amendment And it ensures that copyright “promotes the progress of science” by letting people learn and reuse the ideas expressed.20 Indeed, if there are only a limited number of ways to express a particular idea, copyright denies all protection rather than risk fencing the idea in.21 That system worked well in a world where the expression of ideas was hard, and was therefore the thing copyright needed to encourage. No one should be entitled to own the idea of a painting of a comet appearing over the beach at sunset. But everyone is free to express that idea in their own way, and the variations in how they express the basic idea is the stuff of copyright. The hard work of creation— actually painting the brushstrokes, making the individual choices of color, shading and composition that bring the painting to life—the artist gets to own.22 AI automates much of that formerly hard work. I can ask Dall-E or Midjourney to generate a painting of a comet appearing over the beach at sunset and it will take care of the rest in seconds. If I want more specificity, I can ask it to vary what is in the background, or to change the colors, or to render it in a particular style, and it will adapt the painting accordingly. I have to come up with the basic idea and tell it what I want. But the AI does the bulk of the work that copyright traditionally exists to reward. Dall-E 3 will even turn my basic request into a better prompt in order to give you better images.23 And this isn’t just true of paintings. ChatGPT writes text for you. It will even develop plot devices and characters to fit the parameters of a story you give it. Song AIs are taking off, some generating AI covers of existing songs but others writing entirely new songs that feel and sound like the songs of existing artists—or even ones that feel entirely new. So it’s not just the literal elements of the work, but even many of the non-literal elements of the work, that are contributed by the AI. The result is that, increasingly, the things humans contribute in a collaboration with generative AI will be ideas and high-level concepts. AI will contribute the expression. That turns copyright law upside down. Perhaps it means that the AI should be treated as the copyright author, though as noted above current law doesn’t permit that, and none of the AI companies seem to be pushing for copyright ownership in the expression they generate. Or perhaps it means that no one owns the copyright in AI-generated works: prompters don’t own the ideas they contribute and AIs can’t own the expression they contribute. That might make sense as a policy matter—it is not obvious that generative AI needs the incentive of copyright law—but it is unlikely to be politically sustainable. The instinct Rochelle Dreyfuss has derided as “if value then right”24 is quite strong, and I suspect that as the works AI generates become more valuable people and companies will rush in to claim ownership of them. The most common and most plausible answer, I think, seeks to locate creativity not in the generation of outputs from AI but in the human structuring of the prompts that produce those outputs.25 On this theory, Midjourney or Dall-E might generate an original work of art, but it does so in response to specific instructions from the user. While some instructions are simple, often those instructions are iterated, with users refining the questions they ask ChatGPT/large language models or the directions they give image generation AIs until they get the output they are looking for.26 Sometimes they get it right the first time, but users will sometimes engage in “prompt engineering,” going through dozens of refinements to coax an AI to produce just the right output. And there is strong evidence that the way the prompt is phrased or refined can fundamentally influence the output of foundation models.27 Indeed, companies are now hiring “prompt engineers” with the skill of asking the right questions.28 And “jailbreaking” limits on chatbots by asking questions the right way has become a sport in Silicon Valley.

In analyzing this policy and legal question, there is value in taking a comparative law approach in order to assess how other nations have addressed this important question for the future and present state of copyright law. This article proposes that the United States rule should follow the UK approach. Under the United Kingdom’s Copyright, Designs and Patents Act 1988 (the “CDPA”), there is protection for a category of computer-generated works, that is, those generated by a computer “in circumstances such that there is no human author of the work.”89Section 9(3) of the CDPA provides that the author of a computer-generated literary, dramatic, musical or artistic work, “shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.”90 Although some countries follow the current U.S. approach of declining to protect AI-generated content,91 a number of jurisdictions are finding ways to offer such works copyright protection, including Japan, Ireland, New Zealand, and South Africa.92 It is clear that many common law countries, with more flexible legal frameworks, have anticipated the coming tide of AI growth. It is unclear how the EU and China will eventually resolve this question.93 A rule similar to that in the UK should be adopted in the near future for AI-generated works. This law would provide incentives for the creation and development of AI technology and for the implementation and use of that technology to generate new creative works. This interpretation would make copyright protection for computer-generated works subject to the same requirements that other works must meet. Thus, the work must meet the minimum standard of originality recognized in Feist. It must be independently created by the AI technology (and thus not copied from another source). It must not consist of ineligible subject matter (such as ideas, facts, words, short phrases, scènes à faire, and the like). Copyright protection for AI-generated works would subsist for the same term currently given to works made for hire, which is 95 years from the date of publication or 120 years from the date of creation, whichever comes first. The long-term question—what happens if and more likely when AI technology equals or exceeds human creativity—is no longer merely hypothetical, though it remains unpredictable and futuristic. But it is certainly not beyond the realm of possibility that this situation will arise, and that it will occur more quickly than knowledgeable observers anticipated even a few years ago. If and when it does, one legal solution would be to have the treatment of AI-generated works essentially track the treatment of AI entities themselves. In other words, if they are given legal recognition and status similar to that of human beings, then by right, they would be entitled to the protections of copyright law. In the United States, this result could be accomplished by legislation (likely part of the overall legal status law), by court decisions interpreting the copyright laws, or by an interpretation of the Equal Protection Clause of the Fifth and Fourteenth Amendments. Conclusion In conclusion, the treatment of AI-generated works will continue to vex the courts, the Copyright Office, and Congress. Meanwhile, developments in AI technology will continue to expand the nature and types of works that can be created using AI. Society has already reached the point at which the Turing test—whether AI works are indistinguishable from human creations—has been satisfied. AI-generated content is clearly already as good as—and sometimes better than—many human-produced works. It is, therefore, appropriate for either the courts to interpret the current Copyright Act to encompass AI-generated works or for Congress to enact legislation specifically recognizing their protectability. AI is developing and progressing at such a rapid rate that copyright law will need to adapt and catch up to the reality that creativity is no longer only a human sphere of activity.

In the past few years, the perhaps most striking innovation developments regarding copyright and the creation of new content have taken place with generative AI (Samuelson, 2023). Recent advances in ML have opened up new possibilities to produce literary, musical, and artistic works, and AI has become a new tool to express ideas for hundreds of millions of users. The applications are many and vary across the creative domains. For example, WaveNet was initially developed to help a ML algorithm create seamless artificial voices and has learned to make music itself by analyzing voice material. Current AI systems can also quickly create customized royalty-free background music, jungles, and even entire albums. Also, some video game environments are no longer created by programmers but by the program itself based on predetermined rules and algorithms (e.g., in No Man’s Sky). In visual art, machines can emulate human thought and decide how to alter an image based on a given algorithm. With AI art generators, such as DALL-E, Stable Diffusion, and Midjourney, users can create artwork by entering a prompt, choosing different styles, and further modifying the generated outputs to express their ideas. As the technology is reaching a point where it is difficult to distinguish works created by humans from works generated with AI (Jakesch et al., 2023), the importance of human influence on creative works is put to the test. At present, AI-generated works are not protected by copy- right in most jurisdictions because of the missing human input.“Human authorship” is seen as the guiding principle in copyright law (Hristov, 2016; Ramalho, 2017). Consequently, these works directly enter the public domain. The ongoing development of AI and its ability to make critical, creative decisions either with or without human intervention gives us reason to rethink our understanding of originality and ownership and evolve copyright law. When no protection is granted, the output generated by AI might be considered fair use of the underlying processing and recombination of training data (Sag, 2024). Since the AI is doing much more than copying bits and pieces of existing works and is creating new works based on the techniques learned from studying the copyrighted prior work, much like an art student learns to paint in particular styles by studying existing artwork in that style. However, output substantially similar to original works might still be copyright infringing, giving rise to the technical issue of “memorization” (Sag, 2023a). The case of a song that uses AI-generated voices of the popular artists “Drake & The Weeknd” released in 2023 further highlights a dilemma of online copyright enforcement for platforms. Under pressure from the record label that released the artists’ original works, streaming plat- forms such as Spotify and YouTube removed the song, citing copyright infringement. Given the legal uncertainty, accepting such infringement claims is tricky for Google, which is YouTube’s par- ent organization and a leader in the development of AI. As discussed in Patel (2023), by agreeing with the record label and treating AI-generated music as copyright infringing, Google endangers its own development of generative AI products (which may have much broader applications) under the fair use argument. On the other hand, by not agreeing with the record label, they risk lawsuits and potentially lose necessary licenses for music on YouTube, one of the biggest content categories. Even in countries where AI-generated work is protected by copyright, for example, in the UK, it is not clear who the author of the output is—several eligible personas come into question, such as the architect, the programmer or the user of the model, who have each contributed necessary input for the final work to be created. Depending on whether copyright protection for AI-generated works is granted has further implications on the question of potential copyright infringement regarding the training data, making it more likely that copying original images for training data is legal because the resulting AI software is used to create new, non-infringing artwork

The lanham acts purpose is to guard trademark owner against trademark infringement and unfair competition, and the public against confusion and inaccurate While the unconstitutionality of the disparagement and immoral or scandalous provisions does not affect this purpose, the analyses invoked by the Supreme

Court and U.S. Court of Appeals for the Federal Circuit may. As this article demonstrates, these analyses also apply to several other provisions of the Lanham Act. If challenged, courts may find these other provisions also unconstitutionally violate First Amendment freedom of speech principles. If that occurs, marks capable of infringing or causing confusion will flood the market, severely undermining the stated purpose of the Lanham Act. Courts need to construe statutes narrowly to preserve their constitutionality and, where possible, adopt a saving construction of the provisions of the Lanham Act to prevent this result. Although the Federal Circuit refused to find a saving construction of the immoral or scandalous provision in Brunettia Where possible, courts should in provisions to match already existing judicial analyses. Since application of the obscenity limitation to the immoral or scandalous provision is reasonable, if the government correct. This limitation is necessary not only to protect the marketplace from the proliferation of offensive marks but also to set the tone to save the Lanham Act from an apocalyptic meltdown.

 Aiding the Fight the judicially accepted definition of the phrase, marks depicting obscene material naturally fit the bill. By declaring the immoral or scandalous provision unconstitutional, and refusing to limit it to obscenity, the majority blatantly ignored the definitional match and instead opened the door for the proliferation of discomforting marks. As referenced in Part IV.A supra, though not a point Brunetti presupposes proper application of the immoral or scandalous provision by the USPTO historically. In rejecting the obscenity limitation, the majority relied on the fact that provision involved blasphemous marks touching on religion, which were not obscene At the same time, the majority . determination is made in the context of contemporary attitudes, the concept of what is actually immoral or The fact that examiners used the provision to strike down religious words or symbols during the 1930s,164 a period of religious involvement in the lives of those affected by the Great Depression,165 does not mean this application was proper in and of itself at the time and shouldnt control today. Indeed, in In re Riverbank Canning Co., the first case dealing with the immoral or scandalous provision for rejecting anything, the fact that the USPTO, after rejecting the mark twice, 2008 signals this change.Of course, the movement away from barring registration of religious words or symbols coincides with a shift in contemporary attitudes. majority of [US]PTO rejections under the immoral Similarly, the focus of the cases which the Federal Circuit has reviewed under the immoral or scandalous provision is the same— sex.172 It is difficult to comprehend the provision to obscenity when USPTO examining attorneys, claiming to act on contemporary attitudes, primarily reject marks related to sex. to prevent the registration of even the most patently obscene Such a conclusion is inconsistent with the accepted purpose of the immoral or scandalous provision. As previously mentioned, no legislative history exists to discern the Congressional intent in passing the immoral or scandalous provision.175 Still, however, the principal of legislative inaction suggests Congress consented to the propriety; disgraceful; offensive; disreputable; . . . giving offense to the conscience or moral feelings; . . . or calling out If anything falls within this definition, it is obscene material. By refusing to limit the immoral or scandalous provision to obscenity, the Federal Circuit has exposed the public to potential marks much worse than anything anticipated by the framers of the immoral or scandalous provision. 

The problem with regulations that purport to limit the malicious use of AI is that many of them would also take away the beneficial and benign uses of AI that we see in the creative arts. AI is used in many elements of the creative process already — for instance, the bill’s broad terms could make it difficult to develop AI tools in film editing or make it harder to use existing tools that can lower post-production costs by not having to bring back an individual to re-record lines. Beyond that, the proposal could take away the ability to engage in certain types of content, like parody or sampling. This type of speech has been previously protected by the courts as “fair use,” so a ban on it likely violates the First Amendment. One of the key problems with trying to regulate AI is trying to define it. The definition of artificial intelligence, at its most basic, is the use of programming or machines rather than a living being to solve problems. As a result, the average consumer has been using AI much more frequently and more often than they realize. AI is already in our online searches, the chatbots with which we interact, and the mapping algorithms that help us find the fastest route in traffic. The bill recognizes that there are potential First Amendment concerns, so the text establishes a “First Amendment defense” (something that highlights the proposal’s many problems). An abundance of caution to ensure compliance could result in a chilling effect on existing things like parody TikTok videos, AI-generated political cartoons, or even translating a message to another language if it was in the original speaker or singer’s voice. Reason’s Elizabeth Nolan Brown noted that if this measure becomes law, she expects to see “a lot more takedowns of anything that might come close to being a violation, be it a clip of a Saturday Night Live skit lampooning Trump, a comedic impression of Taylor Swift, or a weird ChatGPT-generated image of Ayn Rand.” “I would also expect to see more platforms institute blanket bans on parody accounts and the like,” Brown added.

In our imperfect information model, the presence of a potentially small number of low-skilled researchers imposes a negative externality on high-skilled scholars, which imposes overpublication as a costly signaling strategy for the latter. In equilibrium, the analysis reveals (1) systematic underinvestment in exploratory research by high-skilled scholars (compared to the first-best perfect information setting), (2) systematic overinvestment in publishing papers (incremental advances), (3) decreased efficiency of resource-based incentives for research, as some of these incentives result in increased publishing, (4) decreased efficiency of reward-based incentives for research, since in the separating equilibrium with output limit, how much time high-skilled scholars dedicate to writing papers depends on the maximum publication output of the low-skilled scholars, and not on the relative reward associated to the two intellectual productions.

Patent-reform activists point out that the number of patent lawsuits has increased by about 60% since 2000, which they cite as evidence that there is a serious problem with the patent system. But does the uptick in lawsuits actually indicate such a problem? In short, no. It might instead reflect a healthy, dynamic economy. Rapid technological advances have spurred more innovation and patents, and courts are now clarifying the nature and boundaries of intellectual property and contract rights. As our colleague Zorina Khan of Bowdoin College showed in a September working paper, the spike in litigation is similar to what happened in the 19th century when the telegraph, telephone and automobile were introduced. Conflict among firms, some of which will naturally spill over into the courts, happens any time disruptive technologies appear. The truth is that patent reform activists have not provided any evidence that the current patent system or Patent Monetization Entities—PME's, the technical, non-pejorative term for a "patent troll"—have hindered innovation and entrepreneurship. To make a convincing case for reform, they would have to prove that the current system is hurting what the public cares about: the quality of the products we buy and the prices we pay for them. Research we conducted with Alexander Galetovic of the Universidad de Los Andes found that innovation rates have been strongest in exactly the industries that patent-reform advocates claim are suffering from "trolls" and a broken patent system. The innovation in these industries is matched with a rapid decline in prices. For example, since 1992, the quality-and-inflation-adjusted price of telephone equipment has fallen by 6.7%, televisions by 14.4% and portable computers by 26.7% a year. Patent-reform activists typically respond to this record with some variation of "yes, but price declines could have been even faster were it not for wasteful litigation." Well, maybe, but thanks in large part to the patent system we have, the current rate of invention in the U.S. might be the fastest in human history. Where is the evidence that society would benefit from undertaking the risky process of reforming a patenting system that has been the envy of the world for more than two centuries? There is one basic reason behind the attacks on trolls: Big Money. Many patent-intensive products—the smartphone in your pocket, the laptop computer in your briefcase—are produced by big corporations that license many patents. The iPhone is a classic example: It contains thousands of patented components, but Apple does not own many of the key ones. It must negotiate for the right to use them. These corporations can make higher profits the less they pay to use patented technology they do not own, and higher profits still by paying nothing at all. The battle over the "right price" for patented technologies takes many forms, one of which is political. Indeed, some corporations are looking to gain a competitive edge by changing the rules of the game. The strategy is to pass patent-reform legislation that weakens the negotiating position of patent holders. Corporations that pay large sums for patented technologies will point to lawsuits, trolls and anything else that will encourage lawmakers to pass such reforms. But when policy makers consider reforming the patent system, they should not rely on often repeated, but never substantiated, claims that patent trolls and lawsuits stymie innovation and the commercialization of complex technologies. They should demand robust evidence that the current system is slowing down innovation. That evidence does not exist.

The Biden administration and Congress have recently made a series of commitments to support industries that are of high strategic importance. The Inflation Reduction Act and Chips and Science Act provide much-needed resources to bolster advances in green energy and increase our domestic supply of semiconductors – two critical long-term priorities. Yet, there’s another significant challenge facing our innovation economy, one that often goes under reported. Abusive patent lawsuits against some of our most innovative companies are on the rise, forcing more and more successful businesses to delay hiring new workers, raising wages, and developing new products. Instead, they’re forced to spend money defending themselves against meritless accusations. The wealthy investors who file these abusive lawsuits, and are often referred to as “patent trolls,” buy up portfolios of broad, unused patents that, in many cases, the U.S. Patent & Trademark Office (USPTO) never should have issued in the first place. They then assert the low-quality patents in lawsuits to accuse others of patent infringement – all this even though the patents being asserted will continue to go unused. The impact on American innovation is devastating. According to one study, each year, patent trolls create $29 billion in direct, out-of-pocket costs from the companies they go after. Another study found that the companies that settle with patent trolls, or lose to them in court, wind up reducing investments in research and development by an average of more than $160 million over the next two years. Massive amounts of money are being drained from the hardworking people who are driving our economy forward to instead line the pockets of wealthy investors who are offering no goods or services of their own. The problem is especially threatening for progress in areas like the renewable energy sector, an industry where products often rely on hundreds or even thousands of underlying patents. U.S. leadership in green technology is critical both to curb emissions and because renewable energy has the potential to be a hub of innovation and strategic advantage for decades to come. The USPTO recognizes the sector’s importance, and its climate change mitigation program will foster research and development in this area. Yet, automakers and other companies generating cutting-edge technology in this sector are constantly forced into battles with patent trolls that drain resources and delay innovation. Letting patent trolls slow down green-energy advancements would be a colossal mistake. Critics say that the problem of patent trolling is an invention of large corporations, and that cracking down on patent trolls will mean hurting “the little guy.” These assertions are not backed up by the data: Analysis has shown that almost 60% of the companies sued by patent trolls are small or medium-sized; patent-troll litigation costs smaller companies more relative to their revenue; and when infringement claims are settled out of court, smaller companies again pay patent trolls more relative to their revenue. It is past time for Washington to fix this broken part of our patent system and ensure innovators and entrepreneurs have the tools they need not just to get by, but to get ahead.

A U.S. trademark registration allows companies to leverage their brand and prevent others from using their name or a confusingly similar name in the U.S. However, even if a trademark is protected in the U.S., it is vulnerable to squatters or “trademark trolls” overseas. Because of the global nature of business today, especially with online commerce and social media, it is important for companies looking to expand internationally to control their trademarks globally and prevent others from squatting on their rights. The last thing a company wants to find out is that it is prevented from expanding its business overseas because it did not fully protect its trademark. While brands are increasingly global, managing and protecting brands globally can be complex. In the U.S., trademark rights are based on actual use. By comparison, trademark rights in many other countries are based on filing. The “first-to-file” system has allowed for a growing trend of “trademark trolls” – companies or individuals who strategically register trademarks belonging to others and then demand large sums of money to sell the trademark back to the original owner or bring an infringement action to enforce their rights as the lawful owner against the original owner. This is exactly what happened to Tesla, Apple, and many others when they sought to expand into China. Therefore, companies planning to expand overseas should consider filing registrations in other countries before someone else does. Specifically, companies seeking further opportunities overseas and are looking to remain in the market long-term should ensure that their brand is protected internationally. Ideally, promptly registering trademarks in a country should be undertaken before any products are sold or manufactured there. However, this is not without complications. Many “first-to-file” jurisdictions allow for the trademark registration to be canceled if not subsequently used in commerce within a certain time frame. Therefore, it is important to identify and prioritize the key markets and understand the jurisdiction’s trademark laws. Although it does take time and resources to register a trademark in different countries, the alternative of exporting goods to another country without a registered trademark leaves the brand unprotected and open to considerable risk.