Two opposed assessments of AI development have crowded out a sober middle ground.
Prominently, modern End Times prophets have ignored real short-term harms and benefits while speculating about improbable future events. This includes both those who believe AI spells doom for the human race and those who believe it will be the saviour of the human race. In the former case, there is the widely publicised resignation of Jacob Coxon, an Anthropic developer who claims that there is a >10% chance that AI kills all humans by 2030 (roughly; how much should we worry about AI killing not all, but only 99% of humans by 2030?). In the latter case, you have people like OpenAI CEO Sam Altman, who believes AI might develop a cure for cancer by the same year. You also have stranger utopians, such as Praxis CEO Dryden Brown, who is buying up land in Uruguay to build an AI-powered utopia — with design modelled on ancient Roman architecture and a caste system inspired by the writings of far-right Italian traditionalist Julius Evola.
On the other hand, certain voices, especially among the left, have rejected these claims as simply hype that’s meant to stimulate their firms’ stock market valuations. Part of this is confirmation bias: we don’t notice AI-generated content when it is done well. Another part, I suspect, is a reluctance to admit the necessity of changing what we consider to be uniquely human. Heated rifts in the literary world have emerged over whether winning submissions for prestigious literary prizes such as the Commonwealth Short Story Prize and France’s Prix du Roman Fnac were generated with AI. People who gloated over the conspicuous and hysterical errors which proliferated in AI image-generation a few years ago have had to swallow their pride.
An emblematic case of this divide is OpenAI’s purported solution to the Navier-Stokes Millennium Prize Problem. Of the seven Millennium Prize Problems — complex mathematical problems unsolved when established by the Clay Mathematics Institute — the problem involving the Navier-Stokes equations has the greatest importance in science and engineering, involving the motion of fluids. On 8 September, OpenAI claimed to have found a counterexample to the problem. OpenAI president Greg Brockman and CEO Sam Altman have both claimed that the firm has achieved artificial general intelligence (AGI): AI software that can outperform humans in any intellectual task.
Immediately after, mathematicians Tristan Buckmaster and Levent Alpöge alleged that their recent advances on the problem had been leaked to OpenAI and integrated into the training data. Many internet commentators then claimed that OpenAI’s purported counterexample was simply stolen and did not constitute a significant advance in artificial intelligence development.
It is probably true that OpenAI employed illicit and immoral means in their pursuit of solving the problem — their staff, according to journalist Karen Hao, have a fixation on achieving ‘superintelligence’ as rapidly as possible.
However, the fact is that OpenAI’s products have gone, in a few years, from failing at rudimentary math problems for high schoolers in GPT-4 to tackling one of the most complex unsolved math problems. Whether Buckmaster and Alpöge’s work was leaked, this demonstrates that new AI models excel in synthesising mathematical information in addition to linguistic structures, which should be sobering for those critics who consistently downplay AI’s capabilities.
The truth is that there is nothing fictional about the dramatic consequences AI will have on the future of the world economy, geopolitical security, and human culture. Already for many communities around the globe today, the consequences could not be more real.
For example, Nairobi, Kenya, had around 40,000 college-educated people in the ghostwriting industry who made a relatively high income by writing essays for overseas students. The rise of ChatGPT wiped this industry out practically overnight. However one feels about the essay-writing gig economy, the spread of AI has wiped out jobs in developing countries which previously exploited the internet for social mobility.
Another, more chilling precursor of things to come is the US Department of Defence initiative Project Maven. The Maven Smart System, developed in league with large firms like Palantir Technologies, Google, and Anthropic, has aided defence officials in targeted airstrikes. One of those airstrikes was aimed at the girls’ school in Minab, killing 120 schoolchildren.
To dismiss the subject of AI as hype and bluff is irresponsible, for it is not as if anyone has any choice to disengage. The data people provide, unwillingly, to large tech companies in their Google searches, published writings, videographed performances, and even daily drives or walks are building the base data for these emerging AI platforms.
At the same time, the fact that AI companies are run by human beings — pursuing profit, comfort, altruism, or whatever other personal motives — ought to impress upon people that apathy and helplessness are precisely the incorrect responses to the current situation.
We must not anthropomorphise AI: Pope Leo XIV made the point well in his encyclical Magnifica Humanitas that AI is not an ‘intelligence’ in the sense of having motivations, desires, internal thoughts, etc. What makes the product so effective is that it feels astonishingly ‘human-like’, but it is for this reason that we should not suppose that AI has any innate bent on annihilating the human race or inaugurating the New Jerusalem. In fact, if anything can cause AI to adopt such ‘motivations’, it is because of the beliefs of its human creators and trainers.
There is absolutely nothing anyone can do to excise AI from the world now that it exists, integrated into virtually every personal device, major computer software, and even military operating systems. However, we have agency over ensuring that AI development takes a different path than it has thus far, where a few clandestine companies run by self-aggrandising CEOs set the agenda. Above all, I believe that a sober reflection on both of these facts demands not indifference and idleness but attention and action.
