Stay up to this point with free updates
Simply log in Artificial intelligence myFT Digest – delivered straight to your inbox.
It seems that the rumors of humanity's demise were greatly exaggerated. Some of our AI overlords, who’re currently flesh-and-blood beings somewhat than robot Terminators, now look like retreating from the industry's more extreme fear-mongering.
Last 12 months, the so-called Doomers, who predicted that a rogue superintelligence could pose an existential threat to humanity, attracted a number of attention. Their alarmism even prompted the British government to host a global AI security summit in Bletchley Park last November. But some top AI leaders are actually dialing down the fear factor and turning up the quantity on hope and hype. Eye-rolling skeptics may note that this renewed burst of techno-optimism coincides with major fundraising by leading AI startups that proceed to pour billions of dollars into developing their models.
Last month, Sam Altman, the co-founder of OpenAI, which just raised one other $6 billion from investors, published an article titled “The Age of Intelligence” by which he predicted that AI would result in amazing achievements: “fixing the climate, establishing an area colony, and discovering all of physics.”
The awarding of two independent Nobel Prizes in Physics and Chemistry to 2 teams, including AI pioneers Geoffrey Hinton and Sir Demis Hassabis, was met with widespread celebration across the industry. Hassabis, co-founder of Google DeepMind, which developed the AlphaFold system that has modeled 200 million protein structures, has spoken enthusiastically about AI enabling science at digital speed.
But perhaps probably the most fascinating demonstration of mood shifting is available in a 15,000-word essay: “Machines of loving grace“, posted this month by Dario Amodei, co-founder of Anthropic. “Fear is a type of motivator, however it is just not enough,” he writes. “We also need hope.”
In his essay, Amodei acknowledges the hazards of AI but focuses on the transformative potential of superintelligence, which he believes could arrive as early as 2026. Such powerful AI, which he likens to a “land of geniuses in a knowledge center,” could significantly speed up progress in lots of areas. These breakthrough innovations would “compress” the twenty first century. “I feel most individuals underestimate how radical the advantages of AI could possibly be,” he writes.
In optimistic mode, Amodei argues that AI could significantly speed up the speed of scientific discovery. This would help us cure many diseases and extend lifespan to 150 years. “AI finance ministers and central bankers” could also distribute global resources more efficiently, helping sub-Saharan Africa increase economic growth by greater than 10 percent. Amodei even speculates that AI could improve societal governance and strengthen democratic institutions somewhat than undermining them.
Amodei accepts that it is simply possible to make “guesses” concerning the future and acknowledges that many readers may view his essay as an “absurd fantasy”. Still, his thoughts are an insightful look into the longer term that AI leaders imagine they’re shaping. Given the astonishing speed of AI development, it will be premature to dismiss such futurism out of hand. In these dark times, we could definitely use a heady shot of optimism.
Yet in some way these AI debates a couple of brilliant future remind me of the theoretical Marxist-Leninist dialectics I spent an excessive amount of time on at university. The early communists believed that vast, impersonal forces would inexorably transform society, almost independently of human intervention. There was no point in resisting the longer term that was coming whether we wanted it or not. But as we all know, history didn't prove that way.
It can be unfair to suggest that Amodei – or his fellow AI advocates – are only as blindly doctrinaire. In fact, they’ve specifically warned concerning the uncertainties and dangers of the technology. But they appear to be committing an analogous teleological fallacy by assuming that humanity will bow to external forces, on this case technology, somewhat than the opposite way around. To paraphrase the Internet, they could need to depart the office more often to “touch the grass” — or at the very least wait for the outcomes of the US election on November fifth.
There is little doubt anymore that AI could bring immeasurable advantages to humanity, however it cannot correct all human imperfections, nor should we wish to. “Nothing straight was ever constructed from the crooked wood of humanity,” the philosopher Immanuel Kant taught us. This lesson applies to all would-be straighteners of humanity: whether or not they are Bolshevik revolutionaries or AI evangelists.
