Stay informed of free updates
Just register at Artificial intelligence Myft Digest – Delivered directly in your reception box.
Two subjects dominated the annual meeting of the World Economic Forum in Davos last week: Donald Trump and artificial intelligence. Of the two, the latter was the most interesting and almost certainly the most important. Great attention in the discussion was devoted to In depthChinese surprise is coming. However, we have simply learned that knowledge is spreading: no country will monopolize these new technologies. It surprised the markets. With new technologies, these “surprises” are not surprising. But that does not change the big question, what the advancement of machine intelligence means for all of us.
Human beings are both social and intelligent. This combination is their “Killer application”. This allowed them to dominate the planet. Human intelligence invented technologies for general ends that have shaped the world, From the tame of fire to the creation of computers. But, with computers who think, it could change. Blaise PascalThe 17th century French mathematician and philosopher, said that “Man is only a reed, the weakest thing in nature, but he is a thinking reed.“Does this uniqueness are coming to an end?”
In Davos, I attended two fascinating discussions on the rewards and risks of advances in AI. One was an interview of Sir Demis HassabisCo-founder of Google Deepmind and joint recipient of the Nobel Prize in Chemistry, by Roula Khalaf, FT editor. The other was a interview of Dario AmodeiFounder and CEO of Anthropic and author of Loving grace machinesby Zanny Minton bededitor -in -chief of The Economist.
Interview with Hassabis highlighted the recent progress of our ability to make a scientific analysis, especially in biology. More than 2 million researchers use AlphafoldHe said, the DeepMind program has developed. “We have folded all the known proteins of science, every 200mn. . . [T]The basic rule is that a doctoral student is needed to find the structure of a protein. Thus, 200mn would have taken a billion years of doctorate. And we just gave all this to the world, for free. This, he developed, is “science at digital speed”. The possibility that has opened up before us is therefore a huge acceleration of medical progress. Indeed, we could have the next 50 to 100 years of normal progress in five to ten years.
In general, argued, we can consider AI as “a country of geniuses in a data centerThe one that the Chinese could have made even cheaper than before. However, these are really geniuses? My test would be to know if, given the knowledge of all physics until 1906, but nothing afterwards, AI would be able to produce the general theory of the relativity of Einstein.
It seems plausible that the impact of such a capacity for problem solving, whether it is a level of “genius” or not, is remarkable. It could, among other things, accelerate knowledge improvements and the growth thus of productivity and the spread of prosperity. Both are desirable. In recent decades, the increase in “total productivity of factors” – the best measure of technical progress – has been modest. In addition, enormous numbers always live in extreme poverty and, depressing, Progress has slowed down.
However, it is also obvious that accelerated progress could also create difficulties. The structure of the labor market can change massively, for example, with, in this case, a sharp drop in the demand of workers whose assets are made, but largely routine, information. Forecasts of such effects vary. A 2023 paper By Erik Brynjolfsson and Gabriel Unger notes that, as was true throughout the computer revolution, the effects on productivity could be modest. However, this time could be different, with arrow productivity, but economic and social changes accordingly and disruptive. Again, according to the reaction of society, the successful AI could lead to a “techno-feudalism”, with even more important concentrations of wealth. The invention of a large number of new treatments could considerably increase health care costs as well as very extensive life management costs even if they are in balance. Are people ready to live alongside their great-great-grandparents? So, apparently, good things could create real challenges.
Beyond that, the development of the envisaged AI creates great risks. How does he control its use by rogue actors, including hostile states, terrorists and mass murderers? What moral judgments do IA make in war? How does he control the use of AI in surveillance? “Big Brother” will he look at us forever? Again, what do we do with the manufacture of false and false news? How does freedom survive all these threats?
Hassabis is clear that we need effective global limits on using AI. At a time of broken international cooperation and contempt for the very idea of ​​an “international order based on rules”, will China and the United States work together in AI security? This seems unlikely, especially because they have different views of the way these technologies must be used.
In 2015, I wrote a skeptic generally article On the (modest) impact probably on the productivity of new technologies. The next few years could finally prove to me the opposite. However, I also noted that if we rather approach “The singularity– Artificial intelligence beyond all human intelligence – everything must change.
One of Frank Herbert’s great ideas Dune The series is that in the distant past (our future) humanity has led a successful jihad against the machines that think. Subsequently, humans had to become superhumans. A main character explain That “humans had put these machines to usurp our sense of beauty, our necessary self, of which we make living judgments. Naturally, the machines were destroyed. »»
This concern could be wise. But I am realistic: AI is out of the Pandora box.
