The right response to AI is more mundane than existential dread

Featured Sponsor

Store	Link	Sample Product
UK Artful Impressions	Premiere Etsy Store

When ChatGPT and other instances of AI software were launched on an unsuspecting public a few months ago, a frenzy of amazement ensued. In its wake has come an avalanche of concerns about where the dizzying developments in software capabilities will take human society, including, surprisingly, people who are very close to the action.

Last month, AI investor Ian Hogarth persisted in the FT’s weekend magazine that “we need to slow down the rush to god-like AI.” A few weeks later, the man dubbed the “godfather” of AI, Geoffrey Hinton, left Google to be able to freely express his concerns, even in a interview with the New York Times. Artificial intelligence professor and entrepreneur Gary Marcus is concerned about “what bad actors can do with these things”. And just today, the FT has a interview with AI pioneer Yoshua Bengio, who fears that AI could “destabilize democracy”. Meanwhile, a large number of investors and AI experts have been doing it called for a “moratorium” on the further development of technology.

Call me naïve, but I found myself unable to let much of the excitement take hold of me. Not because I doubt AI will disrupt the way we live our lives and especially the structures of our economies, of course it will. (Check out this list of the many ways people are they are already starting to use artificial intelligence.) But rather because I struggle to see how even the worst-case scenarios that experts warn us against are qualitatively different from the big problems that humanity has already managed to cause and had to try to solve on its own.

Take Hogarth’s example of an AI chatbot driving someone to suicide. In the XVIII century, reading Goethe THE The Sorrows of Young Werther could presumably have the same effect. Whatever conclusion we should draw, it’s not that AI poses an existential danger.

Or take Hinton, whose “immediate concern is that the Internet will be flooded with fake photos, videos and texts, and the average person will ‘no longer be able to know what is true.'” The inability to see the truth is a fear that seems to be shared by all of the thinkers mentioned above. But lying and manipulation, especially in our democratic processes, are problems that we humans have been perfectly capable of causing without the need for AI. A quick look at some opinions held by large swathes of the US public, for example, shows that (to put it politely) compromised access to the truth is nothing new. And, of course, the ability of generative AI to create deepfakes means we’ll need to become more critical of what we see and hear; and unscrupulous politicians will use the accusation of deepfakes to dismiss damaging revelations about them. But, again, in 2017, Donald Trump didn’t need AI to exist in order to turn the “fake news” accusations against his detractors.

So I think the whiff of existential dread that the latest AI breakthroughs have stirred up is a distraction. Instead, we should be thinking on a much more mundane level. Marcus draws a nice analogy to building codes and electrical standards, and this – rather than an attempt to slow down technological developments themselves – is the floor on which policy discussions should be held.

There are two particularly serious issues (because they are the most actionable) that policymakers should address, especially economic policymakers.

The first is who should be held accountable for decisions made by AI algorithms. It should be easy to accept the principle that we should not allow decisions made by AI that we would not (or would not) allow if they were made by a human decision maker. We’re in poor form on this, of course: We let corporate structures get away with actions we wouldn’t allow individual human beings. But with AI in its infancy, we have an opportunity early on to eliminate possible impunity for real people on the basis of the defense that “AI did it.” (This topic isn’t limited to artificial intelligence, by the way: we should treat dumb computer algorithms the same way.)

Such an approach encourages legislative and regulatory efforts not to get bogged down in the technology itself, but instead to focus on its particular uses and the resulting harms. In most cases, it doesn’t matter much whether a damage is caused by an AI decision or a human decision; what matters is to discourage and penalize the harmful decision. Daniel Dennett exaggerates when he he says in The Atlantic magazine that AI’s ability to create “counterfeit digital people threatens to destroy our civilization.” But he points out well that if the executives of the tech companies developing AI could face jail time for their technology used to facilitate fraud, they will quickly ensure that the software includes signatures that make it easy to detect if we are communicating with an AI .

The law on artificial intelligence be legislated in the EU seems to be taking the right approach: identifying particular uses of AI to ban, restrict or regulate; impose transparency on when AI is used; ensure that rules that apply elsewhere also apply to uses of AI, such as copyright for works of art on which an AI can be trained; and clearly specifying where the liability liesfor example, both with the developer of an AI algorithm and with its users.

The second major issue that policymakers should pay attention to is what the distributional consequences of the productivity gains that AI is expected to ultimately bring will be. Much will depend on intellectual property rights, which ultimately affect who controls access to the technology (and can charge for that access).

Since we don’t know how AI will be used, it’s hard to know how much access to valuable uses would be controlled and monetized. So it’s helpful to think in terms of two extremes. On one side is the fully proprietary world, where the most useful AI will be the intellectual property of companies creating AI technologies. These will be a handful at best due to the huge resources going into creating usable AI. An effective monopoly or oligopoly, they will be able to charge high licensing fees and reap most of the productivity gains that AI can bring.

At the other extreme is the open source world, where AI technology can be used with minimal investment, so that any attempt to restrict access will only result in the creation of a free open source rival. If the author of the leaked Google note “we have no moat”. that’s correct, the open-source world is what we are looking at. Rebecca Gorman of AI alignedclaims the same in a letter to the FT. In that world, the productivity gains from AI will be earned by anyone with the ingenuity or motivation to implement them: tech companies will see their product commoditized and undercut by competitors.

I think it is impossible to know now which extreme we will be closer to, for the simple reason that it is impossible to imagine how AI will be used and therefore exactly what technology will be needed. But I would like to make two observations.

One is to look at the Internet: its protocols are designed to be accessible to all and the language is, of course, open-source. However, that hasn’t stopped big tech companies from trying, and often succeeding, to create “walled gardens” of their products and derive economic income from them. So we should err on the side of worrying that the AI ​​revolution lends itself to the concentration of economic power and rewards.

The second is that where we end up is, in part, the result of the political choices we make today. To push towards an open-source world, governments could legislate to increase transparency and access to the technology developed by tech companies, effectively turning the owner open-source. Among the tools that make sense to consider, especially for mature technologies, large enterprises, or AI instances gaining rapid user adoption, are compulsory licensing (at regulated prices) and compulsory to publish the source code.

After all, the big data that any successful AI will have been trained on is generated by all of us. The public has a strong claim to the fruit of their data work.

Others readable

• “There can be no functioning open trading order without a corresponding security order to underwrite it,” argue Tobias Gehrke and Julian Ringhof, of the European Council on Foreign Relations, in a important analysis how the EU needs to update its thinking on strategic trade policy.

• The digital euro project is steaming ahead but it has yet to win widespread public support.

• The Council of Europe is set up a registry of the damage caused by Russia’s attack on Ukraine. As a formal multilateral initiative, this should make it easier to hold Russia financially accountable for the destruction it has wrought, including through the possible confiscation of his assets.

• The new EU joint procurement platform for natural gas did better than expected in his first race.

Numbers news