Free beta access. 50 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

Particularly, European regulatory standards into US AI policy hampers innovation and development of military AI

Castro and McLaughlin 19

(Daniel, and Michael, “Ten Ways the Precautionary Principle Undermines Progress in Artificial Intelligence”, ) DB

PRECAUTIONARY PRINCIPLE VS. INNOVATION PRINCIPLE

While some people advocate for an almost completely hands-off approach to regulating new technologies, those who recognize that there is a legitimate role for government take two distinct approaches toward action: the precautionary principle and the innovation principle.

The precautionary principle is the idea that if a technological innovation may carry a risk of harming the public or the environment, then those proposing the technology should bear the burden of proving it will not. If they cannot, governments should limit the use of the new technology until proven safe. Those who support the precautionary principle, which calls for government intervention even when there is no clear evidence of tangible and imminent threats of harm, adhere to the cliché it is “better to be safe than sorry.”4 For some technologies, such as nuclear power, the principle makes sense, because the risk of getting it wrong can be catastrophic. However, for most areas of innovation, the precautionary principle leads to more harm than good because it generates hypothetical worst-case scenarios that incorrectly suggest technological advancement presents severe and irreversible threats.5

In contrast, the innovation principle holds that because the overwhelming majority of technological innovations benefit society and pose modest and not irreversible risks, government’s role should be to pave the way for widespread innovation while building guardrails, where necessary, to limit harms. The innovation principle recognizes that market forces, tort law, existing laws and regulations, or light-touch targeted interventions can usually manage the risks new technologies pose. The principle does not, however, argue for a ban on regulation of new technologies. Instead, it advocates for a case-by-case approach, suggesting regulations only in those cases where there is a reasonable expectation that other forces will not suffice and where the potential harms are more than minor. Moreover, in cases where regulations are needed, it stresses the importance of designing regulatory interventions and structuring regulatory enforcement in ways that minimize the harm to innovation, while still achieving the regulatory goals. Finally, it focuses more on ensuring that penalties punish bad actors who cause harm than creating regulations that limit beneficial and benign uses.6 In other words, speculative concerns should not hold back concrete benefits.

Perhaps more so than any government, the U.S. federal government adhered to the innovation principle in its early regulation of the Internet, and this approach fostered a successful era of innovation and growth in the U.S. digital economy.7 In contrast, Europe’s more heavy-handed approach limited and continues to limit digital innovation. For example, many jurisdictions in Europe have restricted the use of ride-sharing apps like Uber because of concerns about the impact on the local taxi industry.8

Given AI’s nascent state of adoption—less than half of businesses worldwide have embedded even one AI-enabled capability into their business process—it is crucial that public policy in all nations spur its development and adoption instead of unnecessarily hindering it.9 Consequently, if policymakers want their nations to achieve the full benefits of AI, they should base their actions on the innovation principle to foster it rather than use the precautionary principle to limit, delay, and constrain its progress.10

Unfortunately, concerns about potential AI harms lead some individuals and groups to advocate for public policies based on the precautionary principle. As a case in point, Elon Musk in 2017 told the world that AI "is a fundamental risk to the existence of civilization” that represents "a rare case where we need to be proactive about regulation instead of reactive."11 He also warned that adopting AI is “summoning the demon” and predicted that these advances could create “an immortal dictator from which we can never escape.”12 Recently Musk has since dialed back his warnings, predicting that AI will not kill us, but only cage us in zoos.13

It is troubling that some people take Musk seriously, but because they do, it is important to rebut such nonsense: Musk is completely wrong. As Max Versace, CEO of the robotics and computing company Neurala and founding director of the Boston University Neuromorphics Lab has explained, “The likelihood of an AI scientist building Skynet is the same as someone accidentally building the space station from Legos.”14 Likewise, University of Washington AI researcher Pedro Domingos has stated that “The Terminator scenario, where a super-AI becomes sentient and subdues mankind with a robot army, has no chance of coming to pass…”15 Unfortunately, the public is often bombarded with hyperbolic and incorrect statements decrying AI, which make it more difficult for policymakers to oppose policies that would hurt AI adoption and to support policies to enable it.

Thus, it is not surprising that several governing bodies embrace the precautionary principle. The European Parliament adopted a resolution in 2017 that research and commercialization of AI and robotics “should be conducted in accordance with the precautionary principle...”16 And Loubna Bouarfa, a member of the European Union High-Level Expert Group on Artificial Intelligence, has even argued that cultural resistance to AI is a “blessing in disguise.”17 After all, if AI is an existential threat to our species, policymakers should be unrelentingly focused on limiting this horror.

Policies based on the precautionary principle are not cost-free propositions, however. In seeking to eliminate potential risks, they can reduce potential benefits and create new problems and unintended consequences.18 For example, some countries have implemented bans on importing or cultivating genetically modified organisms (GMOs)—plants or animals that have altered genetic code—over fears about their safety.19 This is despite a virtually unanimous scientific consensus that GMOs are perfectly safe.20 Bans on GMOs can not only cause higher food prices but also increased greenhouse gas emissions as more forests become farmland to compensate for the lower yields of non-GMO crops.21 Moreover, research suggests GMOs could have saved thousands of lives that perished from malnourishment in African nations that delayed the approval of GMOs.22 Lastly, the ban on GMOs by many European nations has severely limited incomes for many small-scale African farmers.23

Policies based on the precautionary principle almost always stand in the way of innovations that can help the public, and this report identifies 11 policies that would limit the benefits of AI. The remainder of this report provides an overview of AI, lists policies based on the precautionary principle that threaten AI, and analyzes ten detrimental impacts of such policies. To close, it discusses what governments should do to reduce and rectify cases where AI use could be harmful.

WHAT IS ARTIFICIAL INTELLIGENCE?

AI is a field of computer science devoted to creating computer systems that perform operations characteristic of human intelligence, such as learning and decision making. The term does not imply human-level intelligence and the level of intelligence in any implementation of AI can vary greatly. For example, the intelligence level needed for Roomba vacuum cleaners is significantly lower than what is needed for autonomous vehicles.24 Regardless, the development of better hardware, including faster processors and more abundant storage, large data sets, and more capable algorithms in the last decade have helped AI make significant advancements and unlocked new applications.25

AI’s functions include: a) monitoring, such as rapidly analyzing large amounts of data to detect abnormalities and patterns in transactions; b) discovering, including extracting insights from datasets such as the link between a gene and a disease, and through simulations; c) predicting, e.g., using forecasting models to analyze trends to make predictions or recommendations, such as future crop yields; d) interpreting, such as making sense of patterns in unstructured data such as images, video, audio, and text; and e) interacting, both with helping machines interact with one another and also helping humans more easily interact with computer systems.26

There are a vast and diverse array of uses for AI.27 Early adopters include parts manufacturers using AI to invent new metal alloys for 3D printing; pharmaceutical companies using AI to discover lifesaving drugs; mining companies using AI to predict the location of mineral deposits; credit card companies using AI to reduce fraud; and farmers using AI to increase automation. As the technology progresses, AI will continue to bring significant benefits to individuals and societies.

AI is a “general purpose technology,” meaning, among other things, that it will affect most functions in the economy. In some cases, AI will automate work, thereby boosting productivity. By increasing the level of automation in virtually every sector, leading to more efficient processes and higher-quality outputs, AI is poised to boost per-capita incomes. AI can also complete tasks that it is not worth paying a human to do but that still create value, such as writing newspaper articles to summarize Little League games. In other cases, AI adds a layer of analytics that uncovers insights human workers would be incapable of providing. In many cases, it boosts both quality and efficiency. For example, researchers at Stanford have used machine learning techniques to develop software that can analyze lung tissue biopsies faster and more accurately than a top human pathologist can.28 AI is also delivering social benefits, such as rapidly analyzing the deep web to crack down on human trafficking, fighting harassment online, helping development organizations better target impoverished areas, and reducing the influence of gender bias in hiring decisions. Finally, AI will be an increasingly important technology for defense and national security.

AI POLICIES BASED ON THE PRECAUTIONARY PRINCIPLE

Too often policies based on the precautionary principle fail to strike the balance between addressing actual harms posed by AI and not hindering innovation. This failure not only harms the development and adoption of AI but also distracts policymakers from focusing on more important issues, including both legitimate areas of concern and ways in which policy can proactively support the development and adoption of AI. Such misguided policies treat AI in one of three ways: too dangerous to allow (i.e. bans specific uses of AI); too dangerous unless proven safe (i.e. prohibits the technology without special approval from the government); and too dangerous without strict regulatory interventions (i.e. requires the technology to jump through unnecessary and costly hoops before operators can use the technology). These policies are misguided not because they create regulation, but because they create unnecessary barriers to developing and adopting AI due to exaggerated fears of AI or failures to recognize that existing or more nuanced regulation would address potential issues. For example, it is completely legitimate for policymakers to regulate autonomous vehicles to ensure their safe use. But it is another matter for policymakers to limit autonomous vehicles because of possible job losses. We list 11 examples below of unwise policies based on the precautionary principle—that have either become law or have generated support—and we group them into the aforementioned three categories.

Policies That Treat AI as Too Dangerous to Allow

While many critics advocate that the public should fear future uses of AI, or at least carefully plan their use, the most extreme form of the precautionary principle leads to bans on certain uses of AI.29 Various groups and individuals have called for bans on various AI applications, including lethal autonomous weapons, facial recognition, autonomous vehicles, and delivery robots.30 While bans harm innovation and progress, calls for banning new technology have a long history. In the late 19th and early 20th centuries, there were numerous calls to ban automobiles in towns across the United States and Europe. Some individuals lamented the loss of horse-and-carriage jobs, while others complained that automobiles were stirring dust up and causing illnesses. Others called for a ban on automobiles because they opposed the expense of paving roads or because they wanted to preserve the sanctity of the Sunday stroll.31 And in 1982, one New Jersey town even banned pedestrians from using Sony Walkman audio devices “while crossing a street or jogging along a municipal or county thoroughfare.”32 The town created the ban for safety reasons but ignored that individuals could both listen to music and cross streets safely.

In the early 2000s, privacy advocates called for bans of radio frequency identification (RFID) chips, which use radio waves to transmit data, in several use cases, including on government identification documents.33 These advocates warned that stores, governments, and even terrorists would use RFID to track the movements of individuals. For example, the Electronic Frontier Foundation (EFF) argued that a 2005 U.S. State Department proposal to require RFID chips in passports would turn passports into “terrorist beacons,” stating “that's precisely what they'll become if we allow the State Department to move ahead with this plan.34 While the fears of stores, governments, or terrorists tracking individuals with RFID never materialized, RFID tags are helping manufacturers and retailers increase sales and reduce theft and labor costs. They are also in U.S. passports, expediting the scanning of passports.35 Policies that ban technologies do not allow society to gain the technologies’ potential benefits, and most people understand in hindsight that bans only held back progress.

Banning Lethal Autonomous Weapons

Many groups have started movements to ban lethal autonomous weapons—autonomous robotics systems that can independently identify and engage targets based on programmed constraints—due to fears that they will lead to armed conflict on a scale greater and faster than ever before. For example, 116 founders of mostly small robotics and AI companies, including Elon Musk, signed a letter to the United Nations (UN) in 2017 that urges the body to ban lethal autonomous weapons.36 In 2018, the UN Secretary-General António Guterres stated that “machines that have the power and the discretion to take human lives are politically unacceptable, are morally repugnant, and should be banned by international law.”37 Also in 2018, members of the European Parliament adopted a resolution asking member states and the European Council for “the start of international negotiations on a legally binding instrument prohibiting lethal autonomous weapons systems.”38 If policymakers enacted such a ban, it would slow research into AI, as historically, at least in the United States, defense agencies have been a source of significant funding for technology advancement, such as the Internet. And much of the research to support autonomous weapons would yield dual-use technology that could be used for commercial purposes. For example, a fully autonomous tank will likely rely on large portions of the same algorithms and data used to develop a fully autonomous military transport vehicle.39 These same algorithms would be relevant to developing autonomous vehicles for civilian use.

Sign in to recut, update, or save this card.