Culture and Society
PUBLISHED
In early 2026, Peggy Hodgson was vindicated. At least, that’s what various paranormal enthusiasts were claiming. Her story, originally told 50 years earlier, was a famous one: her house, she swore, was haunted by a poltergeist, capable of moving furniture, knocking on the walls, and even possessing her young daughters.
That fame, however, quickly turned to infamy. Skeptics pointed out inconsistencies in both the stories and the proposed mechanics of various claims; the ghostly voice emanating from the children’s mouths, they said, sounded an awful lot like – well, a child; the photo that supposedly showed one girl “levitating” above her bed, cynics said, could just as easily be a picture of her jumping.
Most damningly, video footage seemed to outright show the two girls fabricating pieces of the so-called evidence. And, well, then they admitted it.
It seemed like an open-and-shut case – until AI allegedly proved them all wrong.
In fact, one AI program is said to have declared, the recordings from the house were filled with the voices of ghostly apparitions. In fact, another announced, there was a translucent figure stalking the family photos all along. The “Enfield poltergeist”, despite all the cynicism and skepticism and claims of having made it all up as a prank to scare mum, was real!
It’s enough to make you wonder: wait, ghost hunters are using AI now?
Strange bedfellows
It might seem odd, to connect AI with the paranormal. After all, one is, we’re told, the future: the culmination of computer engineering and cold, scientific progress. The other? A psychological throwback to less enlightened times – a reminder that for all our intelligence and rationality, sometimes we’re still just an easily-spooked ape that forgot how to climb trees.
But that’s what looking from the outside will get you.
The history of the paranormal is deeply enmeshed with the history of science and technology.
Derek Lee
“AI [is] portrayed as rational and objective,” agrees one 2024 paper from a team of researchers at MIT, “which might lead to the assumption that scientifically inclined individuals would find AI predictions more credible.”
But “our findings indicate otherwise,” they write. “Instead, belief in AI predictions was more closely linked to paranormal beliefs”.
For those already inclined to believe in the supernatural, it seems AI is just as legitimate a source of revelation as astrology or messages from beyond. And the more strongly a person believes in the paranormal, the more they trust AI: “Even with entirely fictitious AI predictions, participants rated them as valid, reliable, useful, and personalized,” the team realized.
The trust in technology can run to the extreme. Believe in the paranormal, and you’re also more likely to assume an AI system has agency and experience; when it makes a mistake, you might even ascribe some kind of motivation to it, even if you don’t know why it “felt” the need to “lie” to you.
If all this seems counterintuitive, it shouldn’t. The fact is that AI chatbots may as well have been invented specifically to convince their users that they possess supernatural abilities: they're just human enough, while being just uncanny enough, to ping that bit in our brains that still believes in ghosts and angels.
Generative AI's predictive abilities – in reality just statistical inferences drawn from vast datasets – are basically the tech equivalent of seeing a human figure walk through a wall: it feels impossible, but not so much that you can’t accept it happened.
Even glitches and so-called “hallucinations” can add to the effect, if you’re already inclined to believe it – rather than faulty programming or bad prompts, these become evidence of “an independent, sentient agency that can […] deliver personal messages from beyond the material realm,” writes Samantha Lee Treasure, an independent researcher affiliated with the University of Virginia's Division of Perceptual Studies.
In short – and as strange as it may seem – there’s little reason to expect a conflict between AI and the supernatural. If anything, the opposite is true: the paranormal community is almost perfectly suited to embrace the bold new future of so-called artificial intelligence.
So here’s the question: are they?
Ghosts in the machines
Certainly, it would not be surprising if that were the case. “People often think that science is used exclusively to debunk the supernatural, but that simply isn't true,” says Derek Lee, an Assistant Professor of Literature at Wake Forest University and author of the 2025 book Parascientific Revolutions: The Science and Culture of the Paranormal.
Instead, “the history of the paranormal is deeply enmeshed with the history of science and technology,” he points out. “Science has been continuously used to prove that the paranormal exists or to help us reframe the paranormal in ways that help us wrap our heads around inexplicable phenomena.”
Indeed, outside of the very nerdiest of tech enthusiasts, paranormal communities are often some of the first to embrace new inventions and discoveries – particularly those that might be used to bolster their beliefs in some way.
It’s why Thomas Edison, in the late 1920s, tried to use telephones to commune with the dead; it’s why, 50 years later, ghost-hunters were using home-made EMF sensors and recording electronic voice phenomena (EVP) on the newly available audio cassettes.
“In the 1960s, parapsychologists embraced quantum physics,” points out Lee, “because the search for faster-than-light particles authorized the scientific possibility of precognition.”
Viewed from this perspective, the use of AI in paranormal research is just the latest example in a now long-standing pattern. And as we’ve seen, it’s a particularly evocative one: no mere tool to enable investigation, AI chatbots practically beg to be seen as conduits to other planes of existence.
Here’s the thing, though: AI is, fundamentally, different from those earlier technologies. And for those wanting to prove the existence of supernatural entities, that’s a problem.
Signs from beyond
Listen to an EVP, and perhaps you’ll come away convinced you heard a voice. Dabble with a Ouija board, and you might swear the movement of the glass was more than just the ideomotor response. The conclusions may be shaky – but the evidence, in both cases, was arguably not.
With AI, that’s not always the case. We’ve all seen those wacky statements from chatbots whose algorithms get snagged somehow – claiming that the English Channel has been crossed several times entirely on foot, for example, or that Crungus… you know, exists. But what happens when it starts claiming to be conscious, or commune with ghosts and deities?
It’s no mere hypothetical. Only last month, engineers at Google showed that just a small change to an AI’s programming can produce precisely those effects: it becomes more likely to assert the existence of life after death, belief in a god, and the possession of free will.
Combine that with the link between supernatural belief and some kind of “faith” in AI, and it all adds up to something of a problem for the more traditional paranormal investigators out there.
I think AI has its place and that place will grow rapidly.
Paul Sinclair
“There is […] an interesting question surrounding the use of AI as a possible means of communication with spirits or other entities,” says Paul Sinclair, an independent paranormal researcher, author, and documentary filmmaker who runs the Truth Proof YouTube channel.
“While some people are exploring this idea, I think we need to remain cautious and recognise the difference between an intriguing result and verifiable evidence.”
For Sinclair, AI has some intriguing potential for his research – but not one that’s currently widespread. “From my own experience and knowledge of the paranormal community, AI has not yet become a widely established investigative tool,” he tells IFLScience.
“There are undoubtedly individuals and groups experimenting with it, particularly in relation to audio, video and image analysis, but I'm not aware of any reliable information that would allow me to say how widespread its use actually is.”
Shadowy figures
For all that superstitious laypeople may be taken in by supernatural-themed AI psychosis, then, it seems the professionals are taking a more cautious approach.
“I think AI has its place,” says Sinclair, “and that place will grow rapidly.”
“However, I think it's important to distinguish between using AI to analyse information and assuming that it can identify or establish the cause of something unexplained,” he warns. “Or replace the ability to gauge human emotion when interviewing a witness.”
To be sure, there are those with high hopes for AI’s role in ghost-hunting – but the suggested use cases tend to be surprisingly practical.
Dive into the world of the online paranormal, and you’ll find AI touted for its data collection and analysis; for its ability to recognize patterns or anomalies in the environment too subtle for humans to pick out. And even here, caveats abound: this app is still in beta testing, authors will note; that one adds a filter that can make illusions more likely.
Still, there’s clearly some misunderstanding the limits of AI. Some paranormal researchers suggest that it should be used for enhancing images and video – a fine idea in theory, until you realise you’re instructing the thing that invented Crungus to, essentially, fill your footage with stuff it made up.
Similarly, the idea that AI bots could effectively explore hypotheses concerning the supernatural falls somewhat flat when you remember all the times perfectly reasonable bots started endorsing bizarre and sometimes fatal conspiracy theories.
As strange as it may sound, then, the paranormal community’s stance on the use of AI is perhaps somewhat more sensible than that of the public at large.
“I […] think AI has considerable potential as a tool for paranormal research,” says Sinclair, “particularly when it comes to analysing audio, video, photographic material and identifying patterns that might otherwise be overlooked.”
But it’s important, he adds, to “ensur[e] that human observation, critical thinking and proper investigation remain central to the process.”
Feet on the ground
In the end, the answer to how AI is being used for paranormal research may be disappointingly mundane.
“Where I've seen it used most, is to help create (social) media content,” says Tom Clark, a lecturer in the University of Sheffield’s School of Sociological Studies, Politics and International Relations. “Paranormal social media is now full of AI content in terms of advertising.”
Still, it’s early days. The history of technology in paranormal research is full of squibs – that “spirit phone” never did take off quite like the lightbulb, after all – but many innovations have survived, or even evolved.
Investigators who might once have armed themselves with a navigational compass can now enter sites with sophisticated magnetometers; supposed “hauntings” that relied on reports of cold spots can now be verified with infra-red cameras.
Perhaps, as AI technology improves, it will prove itself useful enough to join the club. “I can see how the advent of AI in the 21st century could […] spur more paranormal inquiry rather than less of it,” says Lee.
The ironic truth, however, is that we simply can’t know the future. “I suspect we're still at a relatively early stage of understanding how AI might be applied to paranormal research,” says Sinclair.
“I’m particularly interested in how AI might develop as a research tool in the future,” he tells IFLScience. “I think [that] is an interesting subject in its own right.”
1