“I don’t know how we’ll respond,” he told Kenny Malone of NPR’s Planet Money. “A hundred years ago, CS Lewis wrote a book called The Screwtape Letters, about a demon in Hell whose job is writing letters to their boss, a senior demon, a manager demon, about their attempts to corrupt a person on Earth. And they have a letter where they say, Wonderful news. My human has just got an invention called the telephone.”

And what’s wrong with the telephone, one might ask? Clark recalls the junior demon explaining, “Whenever my person is alone and would otherwise introspect on themselves and be drawn closer to God, they will instead pick up the phone and call their friends, which brings them closer to Hell. Now, what does that get right and wrong? What it gets right is that things like phones actually massively changed how people relate to one another.”

Clark is certainly right that phones changed us. In 1975, a major fire at a telephone exchange on Manhattan’s Lower East Side disconnected more than 100,000 phone lines — potentially for months, although service was eventually restored after about three weeks. Shortly after that resumption, two media researchers, Alan Wurtzel and Colin Turner, surveyed several hundred of the people who had been affected, and asked them how being without a working phone had felt, and what implications it had had.

Two-thirds of respondents reported that the lack of a telephone left them feeling either “isolated” or “uneasy”, and more than half felt both. Many lamented the loss of the ability to make and receive calls to friends and family. On the plus side, a third of people reported that when the phone lines were down, they visited other people in person more frequently, and the same proportion said they received visitors more often.

There is a complexity here worth observing. As the political scientist Robert Putnam wrote in Bowling Alone (2000), “Socially speaking, the telephone both gives and takes away.” The 1975 fire demonstrated that “the telephone appears to reduce both loneliness and face-to-face socializing”.

Much has changed since 1975, but Putnam’s conclusion remains plausible. The telephone, like all of the hyper-connective technologies that have succeeded it, cuts both ways.

I doubt CS Lewis would have been surprised. The Screwtape Letters are full of descriptions of situations that might equally bring humans closer to God or closer to damnation. Falling in love, explains the senior devil Screwtape, “is simply an occasion which we and the Enemy are both trying to exploit”. The same is true of war. Even sin can bring a person closer to God, by prompting reflection and repentance. If love, war and sin are double-edged swords, we should hardly be shocked to find that the same thing is true of the telephone — or of artificial intelligence.

Why, then, did CS Lewis argue that the telephone was a shortcut to the inferno? Fans of Lewis will have guessed at the answer: he didn’t. Jack Clark had misremembered almost all of his claims. Lewis didn’t write The Screwtape Letters (1942) a century ago; they are not letters from a junior demon, but addressed to him from his uncle, Screwtape. And there is no mention of the telephone at all.

Without wishing to criticise Jack Clark too harshly for misspeaking in an onstage conversation (it’s easily done) there is something magnificently symbolic about hearing such an authority on generative AI produce a string of charming, thought-provoking and believable claims about CS Lewis, all of which turn out to be factually incorrect. Planet Money added a “clarification” to their recording of the conversation, noting that CS Lewis had “talked about the downsides of gadgetry, including the wireless radio” in “a different private letter”.

Curious as to exactly what CS Lewis said about the radio, I tracked down a quotation widely attributed to him — an ironic piece of advice explaining how to avoid God: “Concentrate on money, sex, status, health and (above all) on your own grievances. Keep the radio on. Live in a crowd. Use plenty of sedation.”

But where did he originally write those words? I pasted the quote into Google, but ahead of any search results, Google’s AI Gemini hastened to misinform: “You are quoting or closely paraphrasing CS Lewis from The Screwtape Letters.” Nice try, senior tempter; they were written two decades after Screwtape.

If the telephone can connect us but also drive us away from each other, what should we make of generative AI? I’d suggest that it connects us to — and drives us away from — facts. Anyone who has asked a large language model to produce a research report knows how impressive the results can be — wide-ranging, comprehensive, packed with useful links, and quick. It is a research tool to rival the Google search, and before it, the library catalogue.

And yet with every turn — and now, every Google search — AI confronts us with slick yet forgettable summaries and confident, plausible confabulations. Cheerleaders for AI keep telling me that these fictions hardly ever happen any more. My own lying eyes tell me a different story.

To put it another way, the telephone makes superficial contact easy. That can be the beginning of a more meaningful human relationship, or it can be a replacement for it. Large language models offer the same temptation when it comes to gathering knowledge. A superficial survey becomes effortless. Do we let it catalyse deeper research and understanding? Or replace it?

The choice is ours. Put like that, it verges on being a moral question. We can easily imagine what Screwtape might have made of that.

My dear Wormwood,

I am pleased to hear that the patient has become enamoured of the Machine. Do not try to convince him that it is infallible. That crude approach will not long survive contact with the overwhelming evidence to the contrary.

You must be more subtle. He will learn to distrust it, yes, but his distrust can be kept at a theoretical level. It can easily be curdled into cynicism, an excellent state of affairs from our point of view. Questions about specific facts are what must be avoided. We want him complacent, not curious.

Encourage him to feel that fact-checking is a pedantic inefficiency that is beneath him. Let him enjoy his laziness and call it productivity. Humans have long tended to conflate having an answer with finding the truth. That’s the game.

Your affectionate uncle,

SCREWTAPE

Written for and first published in the Financial Times on 7 October 2026.

Loyal readers might enjoy the book that started it all, The Undercover Economist.

I’ve set up a storefront on Bookshop in the United States and the United Kingdom. Links to Bookshop and Amazon may generate referral fees.

In the late 1970s, Yorkshire miner Chris Campbell sees something that terrifies him. The spirit of a miner who died in a collapse is calling to him. Is the ghostly figure a waking dream, a trick of the light, or something we should take altogether more seriously?

Members of the Cautionary Club have access to all episodes ad-free, monthly bonus episodes, monthly behind-the-scenes video conversations, and our newsletter.

[Apple] [Spotify] [Stitcher] [Patreon]

Further Reading

The primary sources for this episode were: Voices in the Coalshed from the National Coalmining Museum (Chris’s story is here) and The Demon-Haunted World by Carl Sagan.

No surprise, then, to hear Singer’s reaction when he first encountered a prototype sewing machine: “What a devilish machine. You want to do away with the only thing that keeps women quiet: their sewing.” No surprise, that is — except that thanks to some technical improvements and smart business decisions, Singer’s name became synonymous with the devilish machine itself.

In 1860, as Singer sewing machines were starting to catch on, The New York Times opined that “No one invention has brought with it so great a relief for our mothers and daughters . . . Indeed, it is the only invention that can be claimed chiefly for women’s benefit.”

It is certainly true that sewing was thankless work, and a time-consuming duty for many women. And it is also true that, armed with a sewing machine, a seamstress or tailor could make many more clothes in far less time.

One shirt factory in Connecticut in 1860 employed a workforce of 400 people, each with a sewing machine, to produce the same total number of shirts as 2,000 hand stitchers. Output per worker rose from fewer than five shirts a week to two dozen shirts a week. Pay was better, too: $4 a week instead of $3 a week for the hand sewers.

But did Singer’s machine actually liberate women? Much depends on which women — and what we mean by “liberation”.

A new working paper from economists Philipp Ager and Davide Coluccia [pdf] looks at detailed data from 19th-century Massachusetts to try to tease out how the sewing machine changed women’s lives. Their conclusion is that the machine offered working-class women a measure of economic independence — they tended to delay marriage and childbearing and to be more likely to work in paid employment. (Not that these jobs were particularly pleasant: shirt- and dressmaking were the original sweatshop jobs.)

For wealthier women, “liberation” looked different. The sewing machine freed them from a time-consuming domestic duty, which created space in the diary for joining charities and community organisations. But for these upper-middle-class women, the most significant thing the sewing machine made time for was earlier marriage and more children.

It is a story that echoes down the ages: inventions that save labour at the level of a task do not necessarily save labour overall. Instead we “spend” the savings: maybe on more clothes, maybe on more children, maybe in some other way. The canonical example is the washing machine. Given a particular load of clothes to wash, this cuboid miracle-worker saves extraordinary time and trouble. But as Ruth Schwartz Cowan argued in More Work for Mother (1983), the consequence was that clothes would be washed more often. No more removable cuffs and collars, or dark outer layers to hide the grime: just throw everything in the laundry basket every day. Mother will sort it, somehow.

The dishwasher is another “liberation technology”, and older than one might think. Josephine Cochrane patented the first effective dishwashing machine in 1886.

Again, the story of liberation is not quite as straightforward as one might imagine. Cochrane didn’t invent the dishwasher to save herself time in the home; she was wealthy and trying to prevent her servants from chipping her best china.

Although the dishwasher saves only modest amounts of time and effort relative to the washing machine per load washed, it probably saved much more effort per household, because unlike with the washing machine, people saw no need to wash the dishes more often. The term of art here is the elasticity of demand. Demand for clean clothes is elastic: when it becomes much easier to wash clothes, we wash many more clothes. Demand for clean dishes is not: when it becomes much easier to wash dishes, we simply count our blessings.

Queen of the labour-saving technologies, argued Alison Wolf in The XX Factor (2013), was the TV dinner (inventor: bacteriologist Betty Cronin) and more generally all the industrially processed food: the pre-plucked chicken, the pre-chopped salad, the canned soup and jars of sauce, along with the easy availability of affordable takeaway food. It is unglamorous and sometimes unhealthy, but it saved real time.

In the 1960s, prosperous women with university degrees would typically spend four to five hours a day preparing and serving food and cleaning up afterwards. For these women, “putting food on the table” wasn’t a metaphor. It was something that they did quite literally — and it took many hours each week. The American Time Use Survey now suggests that figure has fallen to less than an hour a day for women — and half an hour for men.

The most important “liberation technology” of all, perhaps, was not directly labour-saving in any way: it was the contraceptive pill. Neat, discreet and more reliable in practice than condoms or the diaphragm, the pill gave women greater control over their fertility and therefore a reason to invest in professional careers such as law, medicine and dentistry. The economist and Nobel laureate Claudia Goldin, with Lawrence Katz, has built a compelling case that the availability of the contraceptive pill to unmarried women sparked a generational shift of women into the professions.

Does the history of labour-saving technologies from the sewing machine to the washing machine, the dishwasher to the pill, teach us anything about what we are told will be a revolution in AI? Perhaps. I draw three lessons.

First, these are complicated stories. Even a simple reduction in the cost of stitching led some women to have children earlier, and others to postpone marriage and motherhood. Consequences that make sense in hindsight can be hard to foresee.

Second, elasticity matters. Whether we are talking about clean clothes and clean dishes, we may or may not demand much more when costs fall.

And third, as Isaac Singer’s battered partner Mary Ann could attest, social progress can be advanced by people who have the basest of motives.

Written for and first published in the Financial Times on 30 September 2026.

Loyal readers might enjoy the book that started it all, The Undercover Economist.

I’ve set up a storefront on Bookshop in the United States and the United Kingdom. Links to Bookshop and Amazon may generate referral fees.

In 1612, a young woman encounters a pedlar in the British countryside. A seemingly trivial quarrel spirals into chaos when the pedlar is struck down by an unknown force, leading to one of Britain’s most notorious witch trials.

Members of the Cautionary Club have access to all episodes ad-free, monthly bonus episodes, monthly behind-the-scenes video conversations, and our newsletter.

[Apple] [Spotify] [Stitcher] [Patreon]

Further Reading

Charlotte Meredith’s book is My Mother Is a Witch and This I Know to Be True: The Voices of Pendle. The campaign page, Justice for Witches, can be found on her website. The BBC reported on Nicola Sturgeon’s remarks on Scottish witches and modern accusations of witchcraft, with other coverage from AP News and Reuters. The research on supernatural beliefs can be found on the Yougov (UK) and Gallup (US) websites, and Boris Gershmann’s Witchcraft beliefs around the world: An exploratory analysis.

This script also drew on The Lancashire Witches: Histories and Stories, edited by Robert Poole; The Wonderful Discovery of Witches in the County of Lancaster: Thomas Potts’s original account modernized and introduced by Robert Poole; Something Wicked: The Lives, Crimes and Deaths of the Pendle Witches, by Carol Ann Lee; The Pendle Witch Trials of 1612, by David Holding; The Pendle Witch Trials, by Simon A Williams; and Witchcraft: A history in 13 trials, by Marion Gibson.

But if keeping all those strands in your head is vexing, why not just make up a number and call it P(doom)? The term — shorthand for “probability of doom” — is suddenly mainstream. A couple of weeks ago Evan Hubinger, a senior AI safety researcher at Anthropic, tweeted, “we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade”.

It’s unclear why this particular tweet hit such a nerve. Maybe it’s the timing, after the unsettling LLM-powered hack of the Hugging Face servers. Maybe it’s that maniacally cheery exclamation mark. It’s certainly not the first time such concerns have been expressed by AI insiders. Geoffrey Hinton, neural net pioneer and Nobel laureate, resigned from Google in 2023, citing his age but also warning that the technology was increasingly dangerous. In a recent BBC interview, Hinton suggested that a P(doom) of 10 per cent was “not unreasonable” but added “nobody really knows how to give a sensible estimate”.

Quite so. What would a sensible estimate even look like?

I wrote recently about Philip Tetlock’s work on forecasting. Tetlock realised quite early that many expert prognosticators are extremely vague and rarely held to account. “The economy of Ruritania may soon be facing a currency crisis,” sounds pretty authoritative as a prophecy, but what is the definition of a currency crisis, and does our prophet have a particular deadline in mind?

Tetlock’s solution was twofold: first, pin down the forecast to some objective measure; second, set a deadline and wait. If the prediction is the death of all humans, “set a deadline and wait” is not wholly satisfactory as a tactic.

Tetlock, ever eager to explore the forecasting frontier, is part of a new Automatic AI Risks Outlook (AIRO), in which AI forecasters predict the likelihood that AI will cause various disasters — from the loss of a billion pounds or a thousand lives to the loss of 10 per cent of the world’s population. One advantage of this approach is that smaller losses — of money or of life — are something we can observe, and they might well tell us something about larger risks.

For what it’s worth, those AI forecasters do not foresee our imminent demise, putting the chance of AI killing a billion-ish people by 2030 at less than 0.5 per cent, and less than 5 per cent by 2050.

The chance of catastrophe from any cause (nukes, viruses, asteroid strikes, take your pick) is 8.5 per cent by 2050. You may or may not find this reassuring.

What are these forecasts meant to achieve? Predictions serve many purposes. The most commonplace is small talk — suggesting that Arsenal will win the Champions League or that Kemi Badenoch will lose the leadership of the Conservative Party may serve as a conversation starter. It feels more intelligent than talking about the weather. It’s a kind of cognitive snack food, addictive but hardly nutritious.

Another common reason for making a prediction is to grab attention. The AI merchants certainly have it, distracting us from other risks. Remember when nuclear war was a thing worth worrying about? Terrorism? Fascism? Financial meltdown? Climate change? A deadly pandemic? The most dangerous risks are the ones you stop paying attention to because you’re too busy worrying about something else.

Forecasts are also a good way to boost one’s reputation, especially if they come true — although that is not essential. In the early 20th century, Roger Babson, one of the first economic forecasters, was fascinated by Isaac Newton and believed that Newtonian ideas could be repurposed to make economic predictions. Newtonian forecasting didn’t work especially well, but what did work for Babson was heavily advertising the more successful of his forecasts.

Carissa Véliz’s book Prophecy (2026) offers an example of an attention-grabbing prophecy that turned into something else: a play for survival. King Louis XI’s court astrologer predicted that one of the court’s ladies would die within a week, and then she did. This was a little too accurate for Louis’s tastes — it seemed likely that the soothsayer was either a murderer or genuinely prescient, and neither option felt comfortable. Louis resolved to have the man executed, but first, in a display of regal dark humour, invited the astrologer to predict his own death.

“I will die just three days before Your Majesty,” replied the astrologer, cleverly saving his own skin. It’s a sharp reminder that prophecies are rarely just sincere attempts to predict the future, and sometimes they are not attempts to predict the future at all.

A forecast need not just be small talk, nor a bid for attention or influence. Sometimes predictions really are attempts to help us make better decisions. A data-rich environment can enable useful forecasts even of complex systems, whether it’s the traffic or the weather. Set the parameters, press the button, receive the forecast, act accordingly.

Yet in many cases an accurate forecast is both unlikely and insufficient. In his classic text Scenarios: The Art of Strategic Conversation (1996), Kees van der Heijden describes meeting strategic planners from “a major player in the mainframe computer industry”. They had carefully made forecasts of global demand for computing power, expressed in millions of instructions per second. What they had neglected to do was think seriously about what form that computing power might take in the 1980s and 1990s.

The numerical forecast of demand for computing might have been perfectly accurate, but that was not enough to make better strategic decisions. The real conversation needed to be about whether that computing would power the traditional mainframes the company supplied — or laptops and game consoles. What use is an accurate forecast if it fails to warn you what the future will look like? They should have thrown away their graphs and hired a science-fiction writer.

Which brings us back to P(doom). There are serious conversations to be had about how to regulate AI, how to defend against some foreseeable harms — such as the risk to jobs or the explosion of misinformation — and, yes, how to prevent AI managing to kill everyone on the planet. My question is whether the ritual exchange of P(doom) estimates is a good way to start those conversations, or a good way to avoid them.

Written for and first published in the Financial Times on 23 Sep 2026.

I recently ran the Oxford Half Marathon in October in support of a very good cause and in memory of my mother. If you felt able to contribute something, I’d be extremely grateful.

In 1800s America, something terrifying is feeding on the quiet town of Exeter. As members of George Brown’s family succumb one by one, the town considers drastic measures to protect the living from the dead. 

Members of the Cautionary Club have access to all episodes ad-free, monthly bonus episodes, monthly behind-the-scenes video conversations, and our newsletter.

[Apple] [Spotify] [Stitcher] [Patreon]

Further Reading

Books

Food for the Dead: On the Trail of New England’s Vampires (2013) by Michael Bell.

Everything is Tuberculosis: The History and Persistence of Our Deadliest Infection (2025) by John Green.

Articles

“New England ‘Vampire’ Was Likely a Farmer Named John” by Jason Daley, Smithsonian Magazine (5th August 2019). Available here.

“Old cemetery revealing its secrets” by Jenn Heldt, The Bulletin (24th March 1990). Available here.

“Cemetery Holds Tales of Vampires” by Sam Libby, New York Times (16th February 1922). Available here.

“The Great New England Vampire Panic” by Abigail Tucker, Smithsonian (October 2012). Available here.

“The Man with a Fish on His Back” by Diane Wendy, Distillations Magazine (3rd April 2010). Available here.

“Exeter Hill”, Pautuxet Valley Gleaner (5th February 1887).

“Greene Valley”, Pautuxet Valley Gleaner (5th April 1889).

“Exeter Hill and the Vicinity”, Pautuxet Valley Gleaner (29th May 1891).

“Exeter Hill and Vicinity”, Pautuxet Valley Gleaner (27th November 1891).

“Exeter Hill and Vicnity”, Pawtuxet Valley Gleaner (22nd January 1892).

“Exeter Hill and Vicinity”, Pautuxet Valley Gleaner (26th February 1892).

“Exeter Hill and the Vicinity”, Pautuxet Valley Gleaner (25th October 1892).

“Exeter Hill”, Pautuxet Valley Gleaner (22nd June 1900).

“Exhumed the Bodies: Testing a Horrible Superstition in the Town of Exeter, The Providence Journal (19th March 1892).

“The Vampire Theory: That Search for the Spectral Ghoul in the Exeter Graves”, The Providence Journal (21st March 1892).

“Exeter”, Rhode Island Pendulum (31st July 1891).

“Exeter”, Rhode Island Pendulum (13th May 1892).

“Edwin A. Brown”, The Standard (20th May 1892).


Academic Articles and Historical Surveys

“Historical and Architectural Resources of South Kingstown, Rhode Island: A Preliminary Report”, Rhode Island Historical Preservation Commission (1984). Available here.

“DNA Testing Reveals the Putative Identity of JB55, a 19th Century Vampire Buried in Griswold, Connecticut” by Jennifer Daniels-Higginbotham et al., Genes (Basel), 22nd August 2019, 10(9), p. 636. Available here.

“Bioarcheological and Biocultural evidence for the New England vampire folk belief” by Paul Sledzik and Nicholas Bellantoni, The American Journal of Physical Anthropology, June 1994, 94(2), pp. 269-74. Available here.

“Late-life cardiac interventions and the treatment imperative” by Janet K Shim, Ann J Russ and Sharon R Kaufman, PLOS Medicine, 4 March 2008, 5(3). Available here.

“Death and the treatment imperative: Decision-making in late-stage cancer” by Alexandra Tate, Social Science & Medicine, August 2022, 306. Available here.

Blogs and Websites

“Tuberculosis: A Fashionable Disease?” by Imogen Clarke, Science Museum (24th March 2019). Available here.

“Tuberculosis Part Two: Treatments and Cures”, Science Museum (6th February 2024). Available here.

“Story of Us – T is for Tuberculosis”, Colorado Springs Pioneers Museum. Available here.

“Some Sweet Day, By and By”, Hymnary.org. Available here.

“Have Mercy…” by Jennifer L. Galpern, Rhode Island Historical Society (31st October 2016). Available here.

“Exeter History”, Town of Exeter Rhode Island. Available here.

“Tuberculosis”, Saint John’s Cancer Institute. Available here.

Tetlock’s response was to collect verifiable economic and political predictions from an increasingly wide range of experts, then wait patiently to see which predictions were correct. The result of this exercise was his 2005 book, Expert Political Judgment: How Good Is It? How Can We Know? The subject-matter specialists were barely more predictive than dart-throwing chimps.

Tetlock’s book was a hammer blow to the fragile reputation of forecasters everywhere, but he didn’t give up on the possibility that better forecasts were possible. In Superforecasting (2015), he and his co-author, Dan Gardner, described a select group of genuinely skilled forecasters and highlighted some of the traits and habits that made them effective. With an uncharacteristic touch of hyperbole, Tetlock called these people “superforecasters”. One of the consequences of Tetlock’s work has been that lots of people are now publicly making testable predictions, including some certified “superforecasters”.

So the latest development is — forgive me — predictable. Why not compare some of these published, testable prophecies with those emerging from large language models (LLMs) and other AI systems? For example, do AI systems beat the market at forecasting interest rate decisions? (No, although they’re no worse.)

But asking computers to make these big-picture forecasts is not merely an attempt to test forecasting prowess. It is also, like Tetlock’s original research programme in the late 1980s, an attempt to cut through bullshit and test real skill. Experts and LLMs share a talent for plausible self-justification; a bad prediction, or a good one, sheds new light on those explanations. The title of Tetlock’s 2005 book, after all, was not about forecasting but about expert judgment. One way to evaluate the capability of an AI system is to ask it a question to which nobody — yet — knows the answer.

One effort at this evaluation comes from ForecastBench, a project of the Forecasting Research Institute (president, Philip Tetlock). ForecastBench reports that the results are already impressive, and improving. The typical superforecaster scores 68.9 on ForecastBench’s index of forecasting accuracy. At the time of writing, the best specialised AI forecasters are hitting 68.8. An ordinary human forecast averages 62.8, and regular LLMs with no special tuning are scoring over 61.

Before we ask how seriously to take any of these numbers, we should ask what they mean. One way to understand them is to imagine a forecaster who expresses perfect confidence with every prediction, assigning 100 per cent probability to some and 0 per cent to others, and is correct every time. This infallible oracle would score 100, the maximum score. A forecaster who shrugs and declares that every eventuality is a 50-50 call would hit the chimp benchmark: a score of 50.

The “superforecaster” score of 68.9 would also be achieved by a forecaster who happened to declare a list of events as 68.9 per cent likely to happen — and they all then did — and another set of events as just 31.1 per cent likely, and none of them occurred. Such a benchmark is a lot better than guesswork and a lot worse than omniscience.

Perhaps we should not be impressed that an ordinary LLM is barely worse than a human at predicting events such as “Will there be a presidential election in Ukraine before a ceasefire between it and Russia occurs?” Both the humans and the computers are much closer to the dart-throwing primates than to forecasting perfection.

The fact that specialised AI forecasters now seem indistinguishable from the much-vaunted superforecasters is far more impressive, but there is a long list of caveats. The first is that ForecastBench is not quite comparing like with like: the human superforecasters were answering one set of questions back in 2024, while the AI systems are answering a different set of questions today. This raises a question over the comparison, as clearly some questions are easier than others: “Will Scotland declare independence from the UK by the end of October?” is more straightforward to answer than “Will Vladimir Putin be president of Russia at the end of 2030?”

ForecastBench attempts to adjust for the difficulty of these questions, but admits that the comparison “relies on a statistical extrapolation that grows less reliable over time”. ForecastBench promises that more human forecasts are being gathered, so we may know more soon.

Predictions do not merely describe the future; they are often an attempt to shape it. The mere prospect of highly accurate computerised forecasts, based on systems controlled by the world’s most powerful companies, is not wholly reassuring. “The only way to know for sure where someone will be tomorrow is to make them be there,” says Carissa Véliz, author of Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AI.

But even if AI forecasting systems prove both innocent and accurate, there is something else missing. Forecasts are often valuable not just because of the outcome — a number describing the probability of a future event — but because of the process.

There is an analogy here with using an LLM to write an essay. Regardless of how excellent the result may be, the person who typed the prompt and pushed the button has missed out on the experience of thinking through the process of drafting and redrafting.

So, too, with forecasting. A good scenario-planning exercise is valuable not just because it might produce some thought-provoking descriptions of possible futures, but because it brings together experts and decision makers to ponder the imponderable in a structured manner.

Tetlock himself, with co-authors Barbara Mellers and Hal Arkes, has found that participating in forecasting tournaments tends to reduce political polarisation — presumably because good forecasts require an open mind, the ability to see the world as others see it and a willingness to admit error.

Good-faith forecasting is a practice that makes humans better humans. We should think twice before delegating the task to a machine.

Written for and first published in the Financial Times on 16 Sep 2026.

I ran the Oxford Half Marathon in support of a very good cause and in memory of my mother. If you felt able to contribute something, I’d be extremely grateful.

This episode is available exclusively to Pushkin+ subscribers and members of the Cautionary Club.

[Apple] [Spotify] [Stitcher] [Patreon]

Further Reading

On the Zenith Foundation experiments

Goodfellow, L. D. (1938, reprinted 1992). A psychological interpretation of the results of the Zenith Radio experiments in telepathy. Journal of Experimental Psychology: General, 121(2), 130–144. https://doi.org/10.1037/0096-3445.121.2.130

William Poundstone Rock Breaks Scissors (in the UK, How to Predict the Unpredictable)

Stacy Horn “ESP Games and Academic Politics”

Royal Gazette and Colonist Daily, Thursday October 7 1937

On Louisa and JB Rhine, and ESP Cards

Seymour Mauskopf and Michael McVaugh The Elusive Science (1980)

Feather, S. R. and Ensrud, B. (2018). Joseph Banks Rhine. Psi Encyclopedia. Last updated: 30 April 2026.

Linda Rodriguez McRobbie “How One Man Used a Deck of Cards to Make Parapsychology a Science” Atlas Obscura 27 December 2016

Rhine, J. B. & Rhine, L. E. (1927). One evening’s observation on the Margery mediumship. The Journal of Abnormal and Social Psychology, 21 (4), 401-421

“Doyle Declares Margery Genuine” New York Times 5 March 1927

Massimo Polidoro “A Telepathy Investigation” Skeptical Inquirer Sep/Oct 2018

“Tomorrow, I will be going across channels with zero views, going into their chat, and pumping people up.” But it might not end up quite how you expect… Amygdala, a short film by Joe Edmundson Foot and Stella Harford.

Interested in tabletop roleplaying? The Dice Exploder podcast is my absolute favourite, offering endlessly original, stimulating and creative discussions of different game mechanics.

Caspar Henderson, author of The Book of Barely Imagined Beings, has a substack, The Kinetikon – all about movement, with Caspar’s customary grace and wisdom.

Africa Harford also offers a substack, Elementary, My Dear, exploring the theory and the history of chemistry.

The Runnign Channel have produced a moving documentary about Jess Furness’s attempt to run 42 marathons in a year, in honour of what would have been her late brother’s 42nd birthday. Simple, affecting and inspiring. (Even if I think they have slightly glossed over how hard a single marathon is…)

In 1980s Southern England, intricate crop circles are appearing overnight, and people are desperate for an explanation. Could humans really create something so complex without anyone spotting them? The truth is out there! Actually, the truth is right here. The problem is getting people to believe it.

Members of the Cautionary Club have access to all episodes ad-free, monthly bonus episodes, monthly behind-the-scenes video conversations, and our newsletter.

[Apple] [Spotify] [Stitcher] [Patreon]

Further Reading

This script relied on Rupert Sheldrake’s account of The Crop Circle Making Competition, alongside an article by the Wessex Sceptics and personal correspondence from Matt Ridley; contemporary reporting in outlets such as Time magazine; and four books – Round in Circles: Physicists, Poltergeists, Pranksters And the Secret History of the Cropwatchers, by Jim Schnabel; Circular Evidence: A detailed investigation of the flattened swirled crops phenomenon, by Colin Andrews and Pat Delgado; Crop Circles: Signs of Contact, by Colin Andrews; and Crop Circles: Conclusive Evidence? by Pat Delgado.

Craig Anderson, Mark R. Lepper and Lee Ross describe their research on risk-taking firefighters in Perseverance of Social Theories: The Role of Explanation in the Persistence of Discredited Information, published in the Journal of Personality and Social Psychology. Anderson’s reflections on vaccine hesitancy are found in Psychological processes underlying denial of science-based medical practices, co authored with Sara Prot.