Outsourcing Our Future to For-Profit AI
Within the past month, California Governor Gavin Newsom has vetoed an artificial-intelligence safety bill, and the Royal Swedish Academy of Sciences awarded the Nobel Prize in Chemistry to David Baker, a professor at the University of Washington, and to Demis Hassabis and John M. Jumper, employees of Google’s subsidiary DeepMind and its spin-off Isomorphic Labs.
These two events may seem to have little in common, but, taken together, they suggest that outsourcing humanity’s future to profit-maximizing private corporations is something to be celebrated.
While the California bill was not flawless, it did represent the first substantial effort to hold developers accountable for the potential harms that their AI models might cause. Moreover, it focused not just on any risk but on “critical harm,” like developing weapons of mass destruction or causing at least $500 million worth of damage.
The tech industry, Google among them, lobbied fiercely against the bill, making a very old argument. As the Financial Times editorial board put it, new regulations could “stunt … the emergence of an innovation that could help diagnose diseases, accelerate scientific research, and boost productivity.” Once again, such opportunity costs are deemed more harmful than whatever damage AI might do to people’s ability to control their own destinies, or even to live peacefully in their societies.
The 2024 Nobel marks the first time that the prize has been awarded in a natural science to employees of a multinational corporation. All previous awardees were or had been university professors or researchers at government-funded research institutes, all of whom had published their results in peer-reviewed journals and made their findings available to the world. Whether the Swedish Academy intended it or not, its decision to include the Google researchers helps to legitimize the privatization of science, which is no longer part of humanity’s commons. Like so many resources before it, the science of AI is enclosed within a walled garden accessible only to those who can pay the entrance fee.
True, the AI model AlphaFold2, which earned Hassabis and Jumper the prize, together with its source code, has been made publicly available. According to the AlphaFold.com, “Google DeepMind and EMBL’s European Bioinformatics Institute (EMBL-EBI) have partnered to create AlphaFold DB to make these predictions freely available to the scientific community.” On the other hand, DeepMind holds multiple patents to AlphaFold. According to the logic of property rights, the company, not the public, will always have the final say over the technology’s use. AlphaFold’s website is a “.com,” denoting something fundamentally different from, say, the Human Genome Project, with its “.gov” URL.
In the world of information technology, “free” is never free. Payments are tendered in data, not in dollars. The data that enable AlphaFold to predict the three-dimensional structure of a protein come from the public domain.
DeepMind’s partner in developing AlphaFold is an intergovernmental research organization funded by more than 20 member states of the European Union. According to Jumper, “public data were essential to the development of AlphaFold.” Without data that was compiled and organized by scientists who received taxpayer money for it, there would be no AlphaFold.
Notwithstanding the prescience of public officials in creating this huge database, governments are regularly disparaged for lacking the knowledge, skills, resources, and foresight to promote innovations and advance scientific and economic progress. We are constantly told that only the private sector, with its compelling monetary incentives, can do what it takes to propel the world forward.
In reality, the private sector regularly gets a free ride on work produced by scientists who were supported by public money or employed by public research institutes.
The first satellite was launched by the US government, not Elon Musk; the US military developed the internet before it was commercialized; and pharmaceutical companies rarely invest in basic research. Why bother when you can just wait for scientists funded by the US National Institutes of Health or similar agencies to advance a field to the point where profitable investments can be made?
Such is the logic of profit-seeking corporations. Their purpose is financial returns, not human progress. Once in the game, they seek to monopolize scientific knowledge by securing patents or hiding their findings behind the barriers provided by trade-secrecy law.
Without the helping hand of the state, they would have neither basic science nor legal protections for the monopolies that furnish them with large returns – which they then hold up as proof of their superiority to government.
It is not hard to understand why private companies enjoy this game. The mystery is why governments willingly play into industry’s hands, handing over years of publicly funded research without guaranteeing that the public has its say in determining how it is used. The California legislation would have mandated that AI models include a capacity for full shutdown in case things go wrong, but that provision was killed off with the rest of the bill.
There is nothing new to the argument that if we do not know enough about future damages, we should refrain from interfering in “private” markets, which always perform best without government “interference.”
Oil and gas companies relied on it when denying the risk of climate change and their contribution to it, even as their own research told them otherwise. Yet here we are again. Apparently, we should place our future in the hands of private corporations whose sole purpose is maximizing shareholder value. What could possibly go wrong?
Katharina Pistor, Professor of Comparative Law at Columbia Law School, is the author of The Code of Capital: How the Law Creates Wealth and Inequality (Princeton University Press, 2019).Copyright: Project Syndicate, 2024.
www.project-syndicate.org
Comments
Post a Comment