Government R&D funding solves better than patents

One of history's important lessons is that dramatic, far-reaching innovation often requires significant backing from state actors. 413 A [*1253] version of the Internet, for instance, was first developed by the U.S. Department of Defense before making its commercial debut in the 1980s. 414 Heavy government funding and applied research are also largely responsible for significant innovations in GPS, aviation technologies, space technologies, cellular technologies, nuclear power, nanotechnology, infant formula, bar codes, vehicle tires, microchips, touch screens, vaccines, wind turbines, and many more. 415 In the pharmaceutical sector, government backing is said to be "responsible for 75 per cent of the most innovative new drugs annually." 416

The reason behind this is at least somewhat intuitive: "[T]ruly radical innovation needs patient, long-term, committed finance. This type of finance is hard to find in the short-termist private sector." 417Hence, traditional banks, venture capitalists, and other sources of potential funding are often reluctant to invest in innovations that only promise returns, if at all, after a long period of risky trial and error. 418 In part for this reason, commentators often criticize venture capitalists for primarily investing in projects that promise oversized returns in the near or immediate future. 419 And while such an approach may sometimes prove profitable, it makes it difficult for innovative start-up companies to obtain financing for the type of long-term innovation capable of yielding radical discoveries.

[*1254] Admittedly, the public-private sector debate--that is, whether the government should lead the way, or instead defer to private entrepreneurs, who are often viewed as more capable of innovative activity 420--is an enormous one that this Article cannot hope to settle. Nor need it. Instead, the point here is simply that for truly radical innovation--such as general AI--often both sides are necessary. Hence, we should view government involvement in promoting such radical AI innovation as an important complement to private sector activity, rather than a competitor to or outright replacement for it.

Indeed, a recent review of economic history suggests that for many significant, disruptive innovations, early government support was crucial because clear commercial possibilities were not yet ripe during those early stages of development. 421 Consequently, neither investors nor private sector companies were willing to take on the financial risks necessary to spur on the relevant innovation. 422 By ensuring that the innovations got off the ground, the government thus played a critical role in the early phases of many significant innovations. 423 Later, when commercial applications of the innovations became clearer, the private sector often took the reins, contributing much to the further development of these technologies. 424Indeed, many of the foundational technologies underlying innovations such as Apple's iPhone may have never seen the light of day without early government involvement. 425Apple's later refinements of those technologies certainly provided a number of socially beneficial innovations. 426But without early government [*1255] involvement, society may have never experienced those later contributions. 427

The U.S. government, and governments across the world, remain a significant source of funding for basic research. 428In fact, most of today's AI innovations initially started out in universities. 429 Yet the U.S. government's share of that funding has fallen precipitously in recent years. 430 What is more, its absolute level of funding has also flattened in the last decade, even as basic research needs grow. 431 And constant threats to cut additional funding mean that flattening could eventually nosedive. 432 The lessons of history suggest that if we are to avoid enduring artificial stupidity and make real breakthroughs in achieving general AI, government backing is necessary, and preferably in large doses. 433 This does not mean that private sector entrepreneurs will have no role to play in achieving general AI--they certainly will, and undoubtedly will have much to contribute. But as the history of many significant innovations teaches, often their breakthroughs will only come on the shoulders of governmental involvement. 434 Or in other words, the private sector's [*1256] contributions will only be possible by "building on top of [the] technological infrastructure built by the public sector." 435

Furthermore, the public sector needs greater space to innovate before commercial interests enter the fray. For instance, universities have become so enamored with trying to commercialize their innovations that that commercialization impulse often directs what types of AI innovation university researchers pursue. 436Often, those commercial interests push for narrower, immediately useful forms of AI rather than more general, less predictable ones. 437Hence, greater government funding, with at least some distance from the immediate need to commercialize the funded research activities, seems to be the best way forward to realizing more radical forms of AI innovation.

CONCLUSION

If achieving true AI remains a worthy pursuit, then U.S. innovation policy is in need of a reboot. As this Article has examined, intellectual property rights, particularly patents, are a poor fit for many types of AI development. 438Indeed, for many AI innovators, trade secrecy is often a better choice. 439These intellectual property realities, furthermore, have industrial organization repercussions, as they are likely to contribute to heavy AI industry consolidation. 440That consolidation, in turn, is likely to contribute to enduring artificial stupidity.

But neither strengthening patent rights, nor relying on other forms of law such as antitrust, seem to hold much promise as solutions. Strengthening patent rights is likely to reintroduce a [*1257] number of unintended problems in the AI marketplace that would ultimately tax AI innovators more than help them. 441And antitrust law, with its strict focus on consumer welfare, is unlikely to have much teeth in the AI context. 442Instead, as with many radical innovations before it, achieving general AI will require significant government backing that goes beyond mere lip service. 443 Otherwise, artificial stupidity, though valuable in many contexts, may remain our reality for some time to come.