AI May Very Well Change Our Species

Some technologies increase productivity, but others reshape not only our society but our physiology. Whatever AI turns out to be, the socialist strategy must be the same: increasing the power of labor.

An illustration of prehistoric painters at work in the Lascaux Caves in modern-day France. (Dea Picture Library / De Agostini via Getty Images)

Since the very dawn of humanity, the sudden appearance of new technologies has had profound — yet not immediately obvious — social consequences. As the way we live changes, so do we. Sometimes, profoundly so.

For example, the brains of Homo sapiens everywhere on Earth are smaller now than they were 300,000 years ago. One of several competing explanations of this phenomenon is the theory that the emergence of language and the distribution of knowledge across society reduced the complexity of problems that required solving by individuals. In essence, the first human technologies — culture and language — massively reduced the cognitive burden on individuals. Instead, it was externalized — onto stories, traditions, religions, and crafts. We now have smaller brains than our very earliest ancestors, yet we know much more than they could have ever dreamed of knowing.

Consider another example: the use of pointy sticks, particularly projectiles, in early human societies. This simple technological advancement helped our ancestors hunt large ungulates like mammoths but also may have played a crucial role in fostering a more egalitarian society by diminishing the power of physically dominant males. For famed primatologist Christopher Boehm, this sudden redistribution of the potential for violence explains the decline in the sorts of reactive aggression we see among the other great apes. Pointy sticks reorganized power between humans — the technology fostered an evolved culture of political egalitarianism that strongly distinguishes hominids from the ultraviolent chimpanzee, our closest cousin.

Then around twelve thousand years ago, new technologies began to challenge this “pointy stick” egalitarianism. Humans harnessed the power of evolution and began to selectively breed plants and animals, and this made possible a durable and countable surplus. The so-called Neolithic Revolution wasn’t just about food; it brought new tools, relationships, and social structures. The surplus generated by agriculture paved the way for the establishment of agrarian states. This was when human societies first witnessed the rise of structured hierarchies and the nascent bureaucracies of the state, complete with its trappings of power and subjugation. The weapons that once made possible our egalitarian nature were now the tools of power, exploitation, and domination.

The first experimental horticulturalists were not trying to construct hierarchy out of wild grass seeds. It was an unintended consequence of a very useful innovation. The accumulation of wealth and power, and the state institutions that emerged to defend them, also brought about civilization and written language. As agriculture first emerged, we saw a decline in health and life expectancy; but eventually it facilitated longer, richer, healthier lives, and a much larger human population.

The emergence of capitalism, though, marked a shift in the pace and pattern of technological change. As any student of Karl Marx will know, capitalism is characterized by consistent, and sometimes radical, revolutions in the ways that humans produce what they need. In pre-capitalist modes of social organization, growth was slow and characterized by periodic demographic collapse. In capitalism, output per worker has increased steadily and all Malthusian limits have been overcome.

For the last hundred years, despite huge advancements in technology, certain core features of capitalism have remained stable: the power of the state, dependence on markets, the private appropriation of the social surplus, and so on. But if we are to take the past as our guide, with each new technological advancement, there lies the potential for epoch-defining consequences. The history of humanity is a testament to the transformative power of technology. Past advances have magnified human productive capacities, but some have also resulted in the restructuring of social life and the redistribution of power.

Since at least the Luddites, and consistently since the 1960s, people on the Left — and across the political spectrum, for that matter — have been mostly concerned about the labor market implications of these relentless technological advancements. That is only sensible. New production techniques have often shed labor in order to reduce costs. Luckily, growing total output has in most cases compensated by allowing for the creation of new products and markets.

However, innovation in the digital age has, so far, failed to yield the enormous increases in our total productivity that were anticipated. Computers, robotics, algorithms, internet communications, and now artificial intelligence (AI) based on large language models have all been integrated into the production process. Yet, the growth of per capita productivity is still significantly lower than during the postwar period — particularly in those countries already at the cutting edge of technology.

Since the arrival of ChatGPT, people are again starting to worry. Struggling to come to grips with the implications of recent advances in AI, pundits and politicians have accidentally rediscovered the double-edged nature of technological upheavals. While some foresee a dystopian future where joblessness prevails, and the benefits accrue only to the capital owners, others envision a utopian world free of toil. As with previous rounds of technological advancement, people are beginning to wonder which jobs will be automated and to what extent.

No one can be sure of what the future holds when it comes to such technological breakthroughs. What we have, nonetheless, is a few hundred years of capitalist history, and that does permit a few general lessons to be drawn. The automation of employment has typically resulted in labor being absorbed in other industries. These shifts in the labor force have been associated with meaningful changes in the distribution of power and income across economies. Most of the major innovations of the last century resulted in increasing managerial prerogative and they were intended to do so. Technical change is rarely neutral with respect to the effects it has on the subjective experience of work. The power of working-class institutions, unions, and parties can impact the income and employment effects of automation, but rarely have they shaped the trajectory of technological change itself.

Every now and then, though, a technological breakthrough does, in fact, manage to fundamentally transform the terms upon which we operate — not just as a class or society, but as a species. Like the arrival of language or agriculture, the rise of AI might very well be one of these epoch-defining shifts. But it is not obvious that the effect on employment will be the mechanism through which we experience this upheaval. As we write, AI-enhanced warfare rages in Gaza and in Eastern Europe. New and terrifying modes of surveillance are now being deployed across the planet. And it is increasingly difficult to discern digitally augmented or produced images and sounds from those that were captured from real life. These nonmarket applications are politically meaningful and, frankly, scarier than any employment shifts.

Science and technology are poised to advance in novel ways and might very well progress in directions that are very difficult for many — even all — to understand. With this comes both risk and possibility. For instance, the promise of a healthier, wealthier world is very real, but so is the terrifying buildup of destructive military AI applications.

As was the case since the dawn of the workers’ movement, socialists must engage in politics on a shifting technological landscape. Fighting automation, as such, might be a losing battle — but defending worker autonomy and power needn’t be. Demands for the distribution of the spoils of efficiency gains are the bare minimum. But with respect to the massive unknowns of AI technology, there is no obvious path.

What we do know is that workers and ordinary people ought to have decision-making power in its deployment. In a famous paper, Claus Offe and Helmut Wiesenthal wrote about how collective action problems differ for elites and ordinary working people. Elite interests are transparent — every need is downstream of profit — and this can be achieved through technocrats and lawyers who do their bidding. Working-class interests, however, are never transparent; they always involve dialogue, and must be discovered. Some people simply require more income; others might be older and focused on workplace safety; some have health-related needs or children who require insurance; others still would prefer to bargain for more free time. Dialogue has always been necessary in not only achieving but also in understanding the objectives of ordinary people.

The future of AI is no different, and will require ongoing dialogue to discover what our interests indeed are. This process will be a necessary if not a sufficient condition for the humane governance of new technologies. A decent future will require that the great majority of people have a voice when it comes to the research, development, and deployment of technology. This is only possible with stronger unions and socialist parties contesting power. There is quite a bit of uncertainty around the pace and content of the next few years of technical change. We must be sure to make our mark to the greatest extent possible.

What is good for the few is rarely altogether good for the many, certainly not in the short run. In the long run, let’s hope that — unlike with the arrival of human language and culture — our brains don’t shrink in the process.