LSE’s Asher Kessler and Professor Nick Couldry reflect here on a recent essay by Dario Amodei, CEO of Anthropic, in which he offers a vision of the future and of AI’s role in it. Amodei, who was interviewed this week by the Financial Times, predicts that AI will radically accelerate human progress and alter our world. Is this the future we want?
In October 2024, the CEO of one of the most important artificial intelligence (AI) companies in the world published a 40 page essay in which he imagined the future. Dario Amodei is the CEO of Anthropic, a company he co-founded in 2021 to research and deploy AI in an explicitly safe and steerable way. In Machines of Loving Grace, Amodei predicts that over the next decade, humans maybe able to eliminate most forms of cancer, prevent all infectious disease, and double our lifespans. With the radical power of AI, we can accelerate, according to Amodei, our “biological freedom”; that is, our freedom to overcome the constraints of biology. It is clear that Amodei wants our attention.
The essay starts in a sober, scientific tone, with Amodei distancing himself from Silicon Valley hype about the ‘singularity’ and even the term ‘artificial general intelligence’ (AGI). But that does not stop him developing a very expansive view of how AI will change our lives: across biology and physical health, neuroscience and mental health, economic development, war and peace, and finally work and meaning. Even though he avoids the term AGI, he believes that extremely powerful forms of AI will be with us by 2026.
The result, according to Amodei, is that soon, “a larger fraction of people’s lives” will be spent experiencing “extraordinary moments of revelation, creative inspiration, compassion…”. Harnessing the immediate potential of AI will lead us to drugs that can make every “brain behave a bit better” and more consistently manifest feelings of “love and transcendence”. Alongside ‘biological freedom’ we will gain ‘neurological freedom’ – if, that is, we devolve much of the management of our bodies and minds to AI.
For Amodei, all this is possible, even probable, because AI will do more than add specific innovations: more fundamentally, AI will radically accelerate the rate of progress. In fact, Amodei predicts that over the next five to ten years, we may experience what ordinarily would be 50-100 years of transformation. And here comes his key image: we could be entering a “compressed 21stcentury” of progress.
Yet Amodei acknowledges some limitations. It is less likely,he argues, that global inequality will be reduced, or economic growth will be shared. Nor, even with AI, is our the future one in which democracy or peace islikely to be secured. On the contrary, “the triumph of liberal democracy and political stability is not guaranteed, perhaps not even likely” in Amodei’s AI future.
At this point let’s pause, and ask why in Amodei’s essay certain things are depicted as probable, whilst other phenomena drift out of the realm of possibility. Why, in spite of AI’s extraordinary powers, does it give us a future in which governing through democracy, or living with less inequality, seem less possible than us living until the age of 160? And what does this bifurcation reveal about the ideological assumptions that underlie how Amodei, and other Silicon Valley leaders, imagine the world and our future?
Let’s take the example of democracy and democratic values.
In Amodei’s essay, there is a peculiar relationship to democracy. Yes, some of democracy’s essential functions may be handled better: he even envisions ‘AI finance ministers’. In what seems to be a welcome realism, Amodei anticipates a future in which democracies are less likely to exist, something that – unlike some other Silicon Valley leaders (notably Peter Thiel) – he regrets. But Amodei also stresses how our inefficient democratic governments constrain and limit the true potential of AI. But throughout the essay, there is a complete silence on what democracy entails,and what it means for people’s lives. Democracy is after all, the ability ofpeople to come together and collectively decide on what sort of future world they want to live in. In Amodei’s narrative, democracy and democratic values seem to be erased, or at least ignored, so it is perhaps unsurprising that he sees no reason to be optimistic about their survival. This erasure of democracy’s actual practice is hardly new.
Writing in the 1950s, against the backdrop of the space race,the philosopher Hannah Arendt warned that if we allow science and technology to capture ourability to imagine the future, we will abandon an older faith in collective agency. Whereas previously the future seemed open (in that it was imagined asat least partly the product of open-ended collective decision-making), today the colonization of the future by science and technology seems to have already captured and closed off the future, equating it to never-ending technological breakthroughs under corporate control, rather than what people come to decide in the future. As Satya Nadella, CEO of Microsoft (lead partner of OpenAI, which launched ChatGPT), put it chillingly in his 2017 autobiography: ‘we retrofit for the future’.
Put another way, if (as Silicon Valley seems to demand) we enable the arc of scientific and technological progress to colonize our future, this radically restricts humans from asking perhaps the most important political and social question: “what are we doing (and why)?” Arendt demands that we go on asking this question, which is fundamentally political, not technological:
“The question is only whether we wish to use our newscientific and technical knowledge in this direction, and this question cannot be decided by scientific means; it is a political question of the first order” (The Human Condition, 1958, p.3)
Do we want a future in which some people, almost certainly the richest, almost certainly concentrated in Western countries, double their life expectancy, while others’ life expectancy remains largely unchanged? Do we want a future without democracy? Do we indeed accept a world in which biologists like Amodei (biology is the expertise that he emphasises in his essay, although he also claims an earlier familiarity with neuroscience) have a privileged foresight of a future whose design tools and mechanisms they already control? Should the remarkable calculative feat of AlphaFold in predicting protein structure at inhuman speed (Amodei’s lead example) really dominate our debates about the social benefits and possible harms of AI?
These surely are the questions we can and must ask ourselves. To do so, we must rebuild faith in our agency to take back control of the future that the Big Tech visionaries and oligarchs of the past two decades have captured for themselves.
This post represents the views of the authors and not the position of the Media@LSE blog nor of the London School of Economics and Political Science.