We have to admit that our world has changed radically in recent years, with the advent of the internet and then mobile phones, but we haven't seen anything yet, and with the arrival of artificial intelligence, nothing will ever be the same again, despite all the resistance to change that is destroying the old order.
It has never been clearer that it is not ideas that rule the world, as the era of ideologies would have us believe, when we had to choose between liberalism, communism, and fascism. Instead, it is technological evolution that dictates the rules, as Marx believed, through the advancement of knowledge and technology (knowledge changes the world more than big ideas).
If we have no say in this evolution, what is the point of talking about it? Not being able to decide the future does not imply fatalism or laissez-faire. Rather than fighting futilely over the ideal world we would like to build, which has no chance of ever existing, we must engage in forward thinking to prepare for the world that really awaits us and prevent the disasters that lie ahead. Forward thinking is all the more essential in a time of upheaval like the present, when everything is accelerating, but this is also what makes it almost impossible. There are too many upheavals combining, the effects of which we cannot predict after the fact, any more than we can predict new discoveries or emerging practices. The “hype cycle” is there to show that we are always wrong about new technologies, with science fiction inevitably falling into the worst simplifications when it is taken too seriously. All we ever do is prolong the latest trends, which is very inadequate, but that is our situation—and the question we are faced with remains: what can we do in this context of an uncertain future?
The difficulty of projecting ourselves into the future is obvious when we see the most opposing views on Artificial Intelligence, whose performance has been advancing so rapidly in just a few years and which is only just beginning to sweep into our lives with personal and domestic assistants. While some see nothing new in it, pointing to its current limitations, others fall into the most extreme exaggerations, from fears of being replaced by robots, which will more likely be our partners, to our enslavement by a superior intelligence, if not a mythical Singularity, an exponential extrapolation that makes no sense. The truth is that AI is going to turn everything upside down in the coming years, even more so than digital technology, of which it is the culmination, giving meaning to all the data that is transmitted en masse. We would do well to take this into account, but if we can marvel at the fact that AlphaGo Zero can learn the science of Go in 40 days, which took 3,000 years to develop—just as we marveled at the speed of our computers—this has nothing to do with omniscient intelligence. Kevin Kelly is right to say that the fear of artificial intelligence superior to our own is irrational because intelligence is multidimensional, and superior intelligence has no meaning except in a specialized field, and there are limits to intelligence, which cannot be infinite (or general). Apparently, the only way to approach human capabilities is to take inspiration from our brains, but we don't necessarily have an interest in imitating them completely, as we would also reproduce their madness (an excess of logic) and their risk of error (such as those of group thinking). It should be added that, much more than we think, our intelligence is already largely external (language, books, science, etc.), linked to our historical environment and our education rather than to our brains.
Artificial intelligence will revolutionize many fields, such as medicine, psychiatry, education, (autonomous) transportation, finance, science, justice, etc. The Washington Post has entrusted a program called Heliograf with the automatic writing of more than 800 news briefs covering elections and sports news, and robotic agriculture without humans from sowing to harvesting could further destroy agricultural jobs, etc., to the point where it will be necessary to impose a “human quota” in certain sectors, such as healthcare and nurseries! It is reasonable to think that all this should lead to long-term mass unemployment, although this is not so certain—unemployment being more a feature of our immediate past. A study (by PwC) even claims that artificial intelligence could boost global GDP by 2030. For the United Nations Conference on Trade and Development, while the technological changes underway “cannot explain the current labor market problems,” the problem lies in the fact that “they are occurring at a time of weak macroeconomic conditions in advanced countries.” The risk of robotization is not a catastrophic end of work, but rather an increase in inequality (of capital capture). In any case, while automation has eliminated 140,000 jobs in traditional distribution in the US, it has created 400,000 (lower quality) jobs, and in Germany, where there are eight robots per 1,000 employees, four times more than in the US, it has not led to unemployment but to lower wages. We must nevertheless expect the disappearance of certain professions and considerable job cuts in most sectors, but this will take time and will not mean the end of work, let alone the end of humanity, even if it is AI that will program other AI (which is not so extraordinary). Instead of replacing us, the best strategy is for robots to learn to ask for help or clarification from their interlocutors when they lack the information they need to decide what to do. Google is therefore focusing on the interface between AI and humans with PAIR or People+AI Research, which aims to create empathy and “emotions” in robots to make it easier to work with them. We must not forget that the main advance promised by AI is above all to simplify the user interface, far from eliminating it, until it becomes transparent, and this is what will have the most significant impact (especially when we reach a true understanding of language, which is still very rudimentary). Many of the fantastical fears surrounding artificial intelligence should deflate once the equivalent of a spreadsheet becomes available, allowing these machine learning tools to be used by everyone.
On the other hand, we will certainly see a new radical transformation of work once AI becomes widespread, as with any technological revolution. The economic challenges will probably remain much the same, namely reducing unemployment and inequality, but above all adapting to changes in work, where much will undoubtedly be done from home and on an intermittent basis, with greater emphasis on human development. It is therefore not the end of work but the development of precariousness that will make new protections such as a guaranteed income necessary (according to the Roosevelt Institute, a universal income of $1,000 per month would accelerate growth by 12.56% over eight years).
It is not only work that will be profoundly impacted by the AI revolution, which certainly has its downsides, notably by enabling increased surveillance. A Chinese “smart city” tracks every resident at all times. Above all, AI will impact warfare as fundamentally as nuclear power, and there will be wars between artificial intelligences, such as between stock market robots. It is no coincidence that China is investing heavily in AI in order to become the leader in this field by 2030, including in the military sector. The arms race is only just beginning, as using AI would be the best way to combat malicious AI. There is no need to get carried away and imagine that AI could take over, but we should be concerned about cognitive biases in data and even manipulation. What could help government decision-making and make it less risky could just as easily reinforce authoritarian power. In any case, while we should not overestimate its capabilities in the short term, we must be careful not to underestimate them in the longer term and not believe that it will not make great progress when we are only at the very beginning.
Artificial intelligence will undoubtedly have the greatest impact on us in the years to come, even if we do not really have the means to know how far it will go, but there are many other material processes that will determine our future, which AI can help us to better predict. Apart from technology and climate, demographics will have the greatest impact on our lives in the coming decades, with both aging populations and immigration from Africa, where population growth will be concentrated, leading to widespread intermarriage aided by dating apps (as always, this will be intermarriage, not replacement of the old population). We are also promised custom-made babies, even developing in artificial wombs (assuming that natural gestation is proven to cause too much suffering to the child before and during childbirth!). So, this is far from the end of humanity, even if, between now and then, bioterrorism will be a highly underestimated threat, on a par with nuclear terrorism. Transhumanism, which is very difficult to comprehend, will happen in the longer term, at least to colonize Mars. The decline of the US is ushering in a period of geopolitical uncertainty that could degenerate, but it seems that a form of global governance is nevertheless emerging, constituting a kind of universal state or empire of law that could be described as post-democratic, despite the elections, but which, combined with AI and digital networks, will form a kind of planetary brain. This globalization may also foster separatist sentiments and a revival of local identity. Nor should we forget that many natural disasters can also occur, from the eruption of a supervolcano to a destructive solar flare or a deadly pandemic. The future is never certain, and the only thing that is certain is that it will hardly resemble what we have imagined (either less different or in other ways). Instead of believing that the issue has been settled, that the die is cast and that nothing more will happen, we must remain constantly alert, attentive to what is happening around us, which is always so difficult to assess—a little more intelligence would not go amiss.
