Podcasts
The Last Invention is a sober investigation of the possibility that artificial intelligence will destroy humanity. By Louisa Lim.
The Last Invention confronts the perils of AI
I was not having an existential crisis before listening to The Last Invention. But by the time I finished the fifth episode I was spamming all my friends with the most terrifying page on the whole internet: a list of p(doom) values. This is the likelihood that artificial intelligence will cause a doom scenario, such as human extinction or a new biological weapon, as predicted by some of the most prominent people in the field. The tl;dr is that many of them think we’re doomed.
So it was that I found myself watching a clip of one of AI’s godfathers, Nobel Prize winner Geoffrey Hinton, talking about p(doom) as casually as if he were chatting about the weather, “Because there’s other people who think that 50/50 would be exaggerating it, I think we’ve got a better than even chance of surviving it. But it’s not like there’s only a 1 per cent chance of it taking over. It’s much bigger than that.” A better than even chance of survival? Those are not odds I like. Then I checked the date of Hinton’s statement – June 2024 – and wondered how much his prediction has been changed by the intervening 16 months.
Geoffrey Hinton is one of the star interviewees in The Last Invention, a podcast that I hesitate to recommend on mental health grounds. Nonetheless, it’s the best primer I’ve heard on the development and risks of AI and features a parade of world-class experts for whom the word genius is no overstatement. This eight-part series has shot to the top of the tech podcast charts and its measured tone somehow makes its content even more terrifying.
The Last Invention is hosted by former NPR correspondent Gregory Warner and former New York Times journalist Andy Mills, co-creator of The Daily, who left the institution under a cloud following a scandal over the sourcing of the Caliphate podcast. It is produced and funded by a new outfit named Longview, set up by Mills and Matthew Boll.
The podcast tracks the birth of AI from World War II, when Alan Turing created a giant electromechanical device to crack the Enigma code. The Last Invention’s strength is its ability to explain very complicated technological advances to a general audience without condescension or jargon. A prime example is episode three, which charts AI’s development through its mastery of games, from Deep Blue’s 1997 trouncing of the unbeaten chess grandmaster Garry Kasparov, to a supercomputer’s 2011 conquest of the quiz show Jeopardy, then Space Invaders in 2015, to DeepMind’s 2016 triumph in the ancient Chinese board game Go.
One big takeaway is how often huge advances came from those working on the fringes, derided by the mainstream. One striking moment is when Hinton reminisces about how a colleague warned a keen grad student to keep his distance. “One of the professors in my department told him, ‘Oh don’t work with Hinton. That’ll be the end of your career. It’s a dead end.’ ” The scientific consensus was so disdainful of Hinton’s neural nets that his papers were rejected from machine learning conferences until quite recently.
The podcast’s 10,000-foot view starkly illustrates just how brazenly, and how quickly, the tech titans have done a U-turn on AI’s dangers. As recently as 2023 Elon Musk called AI a civilisational risk, warning a Senate hearing, “There is some chance that is above zero that AI will kill us all.” That same year OpenAI’s Sam Altman, whose company developed ChatGPT, told lawmakers, “My worst fears are that we – the field, the technology, the industry – cause significant harm to the world.”
Two years on, Musk and Altman have both reversed course to become accelerationists, who believe the best way to protect the world from AI is to ensure that they themselves are developing the technology, and as quickly as possible.
Although The Last Invention does a good job of summarising the internecine rivalries motivating today’s ego-driven AI arms race, it lacks voices from the accelerationist camp and the hosts’ “no-comment” approach gives the tech bros a free pass.
One moment of introspection comes in an interview with the second godfather of AI, Yoshua Bengio, who – like Hinton – has become a doomer. Bengio explains the cognitive dissonance that allowed him to continue developing technologies he knew could harm mankind. “It is not a gap due to lack of knowledge,” he says. “It is a psychological barrier. When you work on something, you want to feel good about it. This is a phenomenon well studied in psychology called motivated cognition … What it means is your brain chooses thoughts that make you look good.” Then he makes an admission: “I wasn’t honest with myself at that time. So I regret not having been able to push away my ego, my own self-interests.”
Although The Last Invention is about the most cutting-edge technologies, it sounds more like a throwback to the old-school radio documentary. It has no obvious theme tune, a light touch when it comes to sound design and none of the signposting that has become ubiquitous in the podcast world. The didactic approach of so many podcasts can be irritating, but I was surprised to find that I missed that hand-holding. Signposting in podcasts acts like an aural table of contents, and I felt strangely adrift without any indication of what was to come. Listening to a podcast without any clue of its contents requires a leap of faith that some listeners might not be willing to take. At the same time, I noticed I was listening more intently as I had no idea what would happen next.
The format is a semi-scripted discussion between the two hosts – whose voices are almost indistinguishable from each other – illustrated with expert voices dropped into the conversation. The blokeyness of AI means this podcast fails the Bechdel test: the women it features tend to be journalists or experts rather than players in the field. Its scope so far has been limited to the United States, with scant mention of China’s rapid advances in AI, although that is likely to change in the remaining episodes.
Given my own catastrophising, I was curious how these four months of reporting had affected the podcast hosts. On a call, I asked Warner what his p(doom) now is. Initially, he emphasised the power of human agency, saying, “I think that there will be a point at which some terrible thing happens and people then say, ‘No, we don’t want this future.’ And then that’ll be an interesting kind of catalytic event.” When I pushed him, he was more direct. “I think it’s one. But that doesn’t mean everybody dies.”
This article was first published in the print edition of The Saturday Paper on November 1, 2025 as "Doom machine".
For almost a decade, The Saturday Paper has published Australia’s leading writers and thinkers. We have pursued stories that are ignored elsewhere, covering them with sensitivity and depth. We have done this on refugee policy, on government integrity, on robo-debt, on aged care, on climate change, on the pandemic.
All our journalism is fiercely independent. It relies on the support of readers. By subscribing to The Saturday Paper, you are ensuring that we can continue to produce essential, issue-defining coverage, to dig out stories that take time, to doggedly hold to account politicians and the political class.
There are very few titles that have the freedom and the space to produce journalism like this. In a country with a concentration of media ownership unlike anything else in the world, it is vitally important. Your subscription helps make it possible.
Subscriber exclusive
Send this article to a friend for free.
Share this subscriber exclusive article with a friend or family member using share credits.
