The threat and promise of artificial intelligence – 5/9/2023 – Martin Wolf

In 1900, the UK had 3.3 million horses. These animals provided draft power, transportation, and mounts for the cavalry. Today, only recreational animals remain. Horses are an outdated technology. Their number in the UK has dropped by around 75%.

Could humans too become an outdated technology, replaced by machines that are not only stronger and more dexterous, but also smarter and even more creative?

They say the threat is remote. However, this opinion is a matter of conviction. Perhaps machines can do a lot of what we need better than we can, with the exception of being human and taking care of us like humans do.

However, even if there is no threat of such a revolution, recent advances in artificial intelligence are highly significant. According to Bill Gates, they are the most important development since the personal computer. So what might the implications of that be? Is it possible to control them?

The natural starting point is employment and productivity. An article by David Autor of the Massachusetts Institute of Technology (MIT) and co-authors offers a useful analytical framework and sobering conclusions about what happened in the past. He distinguishes innovation that makes more workers necessary from innovation that automates work. His conclusion is that “the bulk of current employment is found in new job specialties introduced from 1940 onwards”.

But the focus of this new work shifted from middle-wage production and office occupations before 1980 to high-wage professionals and second-rate low-wage service after that. Thus, innovation is increasingly hollowing out middle-income jobs.

Furthermore, innovations generate new types of work only when they complement jobs, not when they replace them. Finally, the effects of automation eroding demand have intensified over the last four decades, while the effect of increased demand has not intensified. None of this is very encouraging, especially since overall productivity growth has been quite modest since 1980.

And what about the future? About this, an analysis by the Goldman Sachs bank is optimistic and at the same time worrying. The study argues that the “combination of significant savings in labor costs, new job creation and growth in the productivity of non-displaced workers raises the possibility of a boom in labor productivity”.

This would be similar to what happened after the appearance of the electric motor and the personal computer. The study estimates that generative artificial intelligence, in particular, can expand annual US labor productivity growth by 1.5 percentage points. The increase would be greater in high-income countries than in developing countries, although the timing of this change is uncertain.

Globally, the study points out that 18% of work could be automated by artificial intelligence, again with greater effects in high-income countries. In the case of the United States, the share of work exposed to artificial intelligence varies between 15% and 35%, depending on the estimate. The most vulnerable jobs will be office and administrative, legal, and architecture and engineering jobs. The least exposed will be construction, installation and maintenance.

Socially, the impact will fall most heavily on relatively highly educated white-collar workers. The danger, therefore, is the downward mobility of the middle and upper-middle classes. The social and political impact of such changes seems all too evident, even if the overall effect is indeed an increase in productivity. Unlike horses, people will not disappear. Furthermore, they have the right to vote.

However, these economic effects are a long way from telling the whole story. Artificial intelligence is a much bigger change than that. It raises profound questions about who we are and what we are. It may be the most transformative technology of all in terms of our sense of self.

Consider some of these broader effects. Yes, perhaps we can have rational, inscrutable judges and better science. But we could also have a world of perfectly faked information, images and identities. We may have more powerful monopolies and plutocrats. We may have almost complete surveillance by governments and corporations.

We may have a much more effective manipulation of the democratic political process. Yuval Harari argues that “Democracy is a conversation, and conversations depend on language. When artificial intelligence hacks language, it can destroy our ability to have meaningful conversations, thus destroying democracy.” MIT’s Daron Acemoglu argues that we need to understand these damages before we let artificial intelligence loose. Geoffrey Hinton, one of the “godfathers” of artificial intelligence, even decided to resign from Google.

The problem with regulating artificial intelligence, however, is that unlike, say, drugs, which have a known target (the human body) and known goals (a cure of some sort), artificial intelligence is a technology. of generic use. It’s multipurpose. It can change economies, national competitiveness, relative power, social relations, politics, education and science. It could change the way we think and create, and maybe even how we understand our place in the world.

There is no way to imagine that we will unravel all these effects. They are very complex. It would be like trying to understand, in the 15th century, the effect that Gutenberg’s printing press would have. We cannot hope to reach agreement on what should be favored and what should be avoided. And even if some countries came to an agreement on this, there would be no way to stop the others. In 1433, the Chinese empire stopped its attempts to project naval power. That didn’t stop other countries from doing it and ultimately defeating China.

Humanity is Doctor Faust. Like the character, she also seeks knowledge and power and is prepared to make almost any bargain to achieve them, regardless of the consequences. Even worse, she’s sort of made up of competing Faustian Doctors, all seeking knowledge and power, as he did. We are experiencing the impact of the social media revolution on our society and politics. Some warn of its consequences for our children. But we cannot escape the bargains we have made. Nor will we stop this revolution. We are Fausts. We are Mephistopheles. The artificial intelligence revolution will continue.

Translated by Paulo Migliacci