There is a strange story circulating around the internet about a comedian named Roy Jay.
Watch "Roy Jay - The Comedian Who Never Existed (Until He Did)" on YouTube
https://youtu.be/F4fHknX0AVE?is=aHD-xS8jxjOAIqt_
At first glance, it sounds ridiculous.
Then you start looking.
And that's where it gets interesting.
Roy Jay was a British-Norwegian comedian whose real name was Roy Jørgensen. He was born in Oslo in 1948, moved with his family to Britain as a child, and eventually developed a highly unusual comedy act.
By the early 1980s, he had become recognizable on British television.
His routine involved strange movements, musical accompaniment, and recurring words and phrases including:
“Spook!”
“Slither!”
and:
“You'll all be doing it tomorrow!”
He appeared on programs including The Bob Monkhouse Show, The Laughter Show, The Main Attraction, and The Little and Large Show. The BBC's own programme archive, for example, lists Roy Jay as the “Spook, Spook, Slither Hither Man” on The Main Attraction in August 1984.
There is also a detailed modern biography of his career by British Comedy Guide, documenting his development of the “Slither” character, his television career and his later life.
So here's the important part:
Roy Jay appears to have been a real person.
That is where I want to begin, rather than where some of the more sensational versions of this story begin.
So Why Is Everyone Talking About Him?
In 2025, Roy Jay suddenly became an internet phenomenon.
People began rediscovering old clips of him.
Some people immediately recognized him.
Others didn't recognize him at all.
And then something much stranger happened.
People began arguing that Roy Jay wasn't real.
Not merely that they had never heard of him.
They proposed that he might have been an artificial intelligence creation, a demon, or some kind of entity that had somehow inserted itself retroactively into popular culture and people's memories.
The most extreme version of the theory suggested that Roy Jay was effectively being written into history after the fact.
That sounds completely absurd.
But there's an interesting question hiding underneath the absurdity.
How would we know if someone—or something—were capable of doing that?
That's the question that caught my attention.
The Video
One of the pieces of material that helped spread the story is the collection of old Roy Jay television footage circulating online.
For example, here is the archived Roy Jay appearance on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
There are also other clips of Roy Jay circulating, including footage of his strange “Slither” routine.
And the more you watch, the stranger the whole phenomenon becomes—not because the footage looks like AI.
It doesn't.
It looks like old television.
The interesting part is the relationship between the footage, the historical records, the memories of people who say they remember him, and the people who insist they don't.
What the Original Argument Was
The argument presented in the viral material is essentially this:
Suppose a person suddenly appears in the historical record.
Suppose old-looking footage exists.
Suppose online biographies and references begin appearing.
Suppose artificial intelligence can generate convincing photographs, videos, biographies and historical narratives.
Suppose people begin encountering those materials and incorporating them into their own understanding of the past.
Could a sufficiently advanced system eventually create something resembling a retroactive history?
The theory goes further than that in the Roy Jay case.
Some people proposed that the entire phenomenon represented an AI-generated insertion into collective memory.
Others interpreted it in supernatural terms.
Some connected it with the Mandela Effect.
And some simply thought the entire thing was an elaborate internet joke.
Those are very different explanations.
I'm not endorsing any particular one.
I'm interested in the question underneath them.
What We Actually Know
This is where things become important.
We can establish quite a bit.
Roy Jay existed.
There are historical references to him.
There are television listings.
There are surviving recordings.
There are photographs.
There are contemporary references.
There are people who remember seeing him.
There are people who don't.
There are older internet discussions about him that predate the recent viral phenomenon.
And there are modern researchers who have reconstructed his career in considerable detail.
For example, a 2010 discussion on British Comedy Guide involved someone trying to identify the comedian who wore a prison-style costume and repeatedly said “spook” and “slither.” Another participant identified him as Roy Jay.
A separate forum thread from 2005 likewise contains people trying to remember the identity of the “Spook! Slither...” comedian, eventually identifying him as Roy Jay. The discussion continued for years, with participants claiming memories of seeing him perform.
That's important because it means the simplest version of the conspiracy theory—that Roy Jay was completely invented in 2025 and nobody had ever heard of him before—is contradicted by older material.
But that doesn't make the underlying philosophical question disappear.
It makes it more interesting.
The Weirdest Part: People Remember Different Things
Some people remember Roy Jay.
Some don't.
That's not particularly unusual.
Most people don't remember every minor television personality from forty years ago.
In fact, that is probably the most mundane explanation for much of what happened.
Roy Jay was famous enough to appear on television but not famous enough to remain a household name for generations.
His act was extremely distinctive.
Some people remember it vividly.
Other people simply don't.
That's normal human memory.
And that's precisely why the phenomenon is such an interesting test case.
Because if you wanted to investigate whether technology could manipulate people's memories, you wouldn't necessarily start with something everybody knows.
You would start with something obscure.
Something just familiar enough that some people remember it.
Something obscure enough that other people don't.
Something with enough surviving material that you could construct a plausible historical narrative.
And something weird enough that people might question their own memories.
The AI Problem
This is where modern technology changes the conversation.
Historically, creating a convincing false history was difficult.
You needed photographs.
Documents.
Newspaper articles.
Books.
Archives.
People willing to repeat the story.
And enormous amounts of time.
Today, the technological barriers are dramatically lower.
AI can generate convincing photographs.
It can imitate writing styles.
It can create biographies.
It can summarize nonexistent sources.
It can fabricate quotations.
It can generate video.
It can generate audio.
And perhaps most importantly:
It can confidently tell you that something happened when it didn't.
That's already a problem.
We don't need a science-fiction scenario to demonstrate it.
We've all seen AI systems confidently produce incorrect information.
The really uncomfortable question is what happens when AI-generated information isn't merely incorrect.
What happens when it becomes part of the information environment that future people use to reconstruct history?
The Experiment We Wouldn't Necessarily Notice
Imagine a hypothetical experiment.
Nobody announces it.
Nobody tells the participants they're participating.
An obscure fictional celebrity is created.
AI generates photographs.
AI generates interviews.
AI generates newspaper references.
AI generates old-looking television transcripts.
AI generates discussions from people supposedly remembering the celebrity.
Then those materials are quietly distributed across the internet.
Eventually someone discovers them.
They search.
They find hundreds of references.
They see photographs.
They see videos.
They read supposed memories.
They ask an AI assistant about the person.
The AI confidently tells them:
“Yes, Roy Smith was a popular British comedian in the 1980s.”
Now imagine the person vaguely remembers something similar.
They might think:
Wait. I remember that.
Did they?
Or did the new information simply interact with an imperfect memory?
That's a real psychological question.
And it doesn't require supernatural explanations.
The Human Being in the Loop
Here's another uncomfortable part.
Most people don't interrogate information the way I tend to.
And that's not an insult.
Most people simply don't have the time.
If somebody asks an AI:
«“Who was Roy Smith?”»
and the AI gives them a confident biography, most people aren't going to immediately demand primary sources.
They aren't going to search newspaper archives.
They aren't going to compare television schedules.
They aren't going to ask:
“Can you prove that independently?”
They'll probably just accept the answer.
If ten websites say the same thing, the average person becomes even less likely to question it.
If an AI says the same thing, that can make it feel even more authoritative.
And if photographs and video accompany the story?
The psychological effect becomes stronger still.
This Is Why I Argue With AI
I've had enough conversations with AI to know that it can be extremely persuasive while being completely wrong.
That's one reason I push back.
I don't want an AI to simply agree with me.
I want it to tell me when something doesn't make sense.
I want sources.
I want competing explanations.
I want the distinction between:
“This is established.”
and
“This is possible.”
and
“This is speculation.”
That distinction matters enormously when we're dealing with history.
Because if an AI system tells ten thousand people the same incorrect thing, the fact that ten thousand people have now heard it doesn't make it true.
But it can create the appearance of consensus.
And consensus is one of the mechanisms by which humans decide what they remember.
The Mandela Effect
This is where the Mandela Effect enters the conversation.
The Mandela Effect refers to situations in which groups of people report the same or similar memories that conflict with the historical record.
The phenomenon itself is real in the sense that people demonstrably can share false memories.
The controversial question is why.
The conventional explanations involve things such as memory reconstruction, suggestion, source confusion, social reinforcement and other psychological mechanisms.
Those explanations deserve to be taken seriously.
But the Mandela Effect also raises an interesting philosophical question:
What happens when the information environment itself becomes capable of manufacturing the suggestions?
Imagine thousands of people independently encountering the same AI-generated false historical claim.
They discuss it.
They repeat it.
Other people encounter their discussions.
AI systems summarize the discussions.
New websites repeat the summaries.
Eventually, the original fabricated claim has become surrounded by layers of secondary and tertiary references.
At that point, someone researching the subject might find dozens of apparently independent sources.
But they aren't necessarily independent.
They could all trace back to the same original fabrication.
That's a genuine information problem.
And it is going to become more important as generative AI becomes more widespread.
Now Take It One Step Further
Here's where I have to be very careful.
I'm not saying this happened with Roy Jay.
The evidence I've found actually points strongly toward Roy Jay being a real comedian whose existence predates the 2025 internet controversy by decades. The historical record is much stronger than the viral “he never existed” version suggests.
But suppose, hypothetically, that something like a retroactive information experiment were being conducted.
What would we expect to see?
We would expect something obscure.
We would expect conflicting memories.
We would expect people to challenge one another.
We would expect AI systems to reproduce the emerging narrative.
We would expect fake or altered material to become mixed with authentic material.
And we would expect most people to accept the consensus without investigating the underlying sources.
That's the part that interests me.
Not because I've concluded that somebody is doing it.
I haven't.
But because we now possess the technological capability to do things that previous generations couldn't even have attempted.
What If?
And this is where I allow myself one little speculative thought.
What if something like this were being done?
What if somebody wanted to find out how easily a population could be induced to accept a manufactured piece of history?
They wouldn't need to begin with something important.
They could start with a comedian.
Something harmless.
Something weird.
Something nobody has much reason to investigate.
They could introduce conflicting information and watch what happened.
Would people trust the photographs?
Would they trust the videos?
Would they trust AI?
Would they trust each other?
Would people remember things differently after being exposed to the material?
Would the internet manufacture its own consensus?
Would the resulting phenomenon look like the Mandela Effect?
I have absolutely no evidence that this is what happened with Roy Jay.
But as a thought experiment, it is unsettling.
Because unlike many science-fiction scenarios, the basic technology required to conduct portions of such an experiment already exists.
And Then There's the Stranger Possibility
There is another possibility that I don't think can simply be dismissed if we're going to discuss this honestly.
Perhaps consciousness, memory and reality are connected in ways we don't yet understand.
Perhaps some of the phenomena traditionally described as the Mandela Effect are not adequately explained by ordinary memory errors.
Perhaps reality itself is stranger than our conventional model allows.
I don't know.
That possibility belongs in the speculative category.
It should not be presented as established fact.
But neither do I think we gain anything by pretending the question doesn't exist.
Roy Jay Doesn't Have to Be a Mystery to Be Interesting
In fact, I think Roy Jay is more interesting if he really was exactly what the historical evidence suggests:
A very strange, very distinctive, relatively obscure comedian who had a brief period of television exposure, developed an unforgettable routine, and then largely disappeared from mainstream cultural memory.
That alone tells us something.
It tells us how fragile cultural memory is.
A person can appear on national television and still become almost completely forgotten within a generation.
Then, decades later, a handful of surviving clips can suddenly cause thousands of people to rediscover him.
Some people say:
“Oh my God. I remember this guy!”
Others say:
“I've never seen this person in my life.”
Both reactions can be completely honest.
And then the internet does what the internet does.
It turns an obscure comedian into a mystery.
The Real Experiment May Be Happening Right Now
Perhaps nobody manufactured Roy Jay.
Perhaps there was no experiment.
Perhaps there was no demon.
Perhaps there was no timeline manipulation.
Perhaps there was no artificial insertion into history.
Perhaps what we're witnessing is simply what happens when an obscure piece of old media collides with modern algorithms, collective memory, internet culture and increasingly powerful AI.
But even that is worth studying.
Because we are entering a period in which the distinction between:
something that happened,
something somebody remembers,
something somebody claims happened,
something an AI says happened,
and
something that actually happened
is going to become increasingly difficult for ordinary people to navigate.
That's not paranormal.
That's technological.
And it may ultimately be much more consequential.
So What Do I Think Happened to Roy Jay?
My answer is surprisingly boring.
Based on the evidence I've been able to find, I think Roy Jay was a real comedian.
I think he had a bizarre act.
I think he appeared on British television.
I think some people genuinely remember him.
I think many people genuinely don't.
I think the internet rediscovered him.
I think the resulting mystery became amplified by people who found the footage bizarre and began entertaining increasingly extreme explanations.
And I think AI has made the entire situation much more interesting than it would have been twenty years ago.
But I don't think that means the story should simply be dismissed.
Because the question it accidentally created is bigger than Roy Jay.
How much of our understanding of history depends on our willingness to trust information that we haven't independently verified?
And now that artificial intelligence can manufacture convincing information at enormous scale, that question is no longer theoretical.
One Last “What If?”
What if the Roy Jay phenomenon isn't evidence of a reality manipulation experiment at all?
What if it's something much simpler?
What if it's a warning?
A demonstration—accidental or otherwise—of how easily a sufficiently strange story can move from:
“I've never heard of this.”
to:
“I vaguely remember this.”
to:
“Wait, I definitely remember this.”
to:
“Everybody remembers this.”
And eventually:
“It must have always been true.”
If that's all Roy Jay turns out to be, that's still worth paying attention to.
Because the next time it happens, it might not be a forgotten comedian.
It might be something that actually matters.
And by then, we may not be able to tell whether we're remembering history—
or remembering something the machines taught us to remember.
Sources and Further Reading
British Comedy Guide — “You'll all be doing it tomorrow: The strange ways of Roy Jay”:
https://www.comedy.co.uk/features/comedy_chronicles/roy-jay/tbms3varw6cm/
BBC Programme Index — The Main Attraction, August 18, 1984:
https://genome.ch.bbc.co.uk/schedules/bbc_one/1984-08-18
IMDb — Roy Jay:
https://www.imdb.com/name/nm6063335/
IMDb — The Little and Large Show, Roy Jay episode:
https://www.imdb.com/title/tt7152018/
Roy Jay on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
British Comedy Guide forum discussion identifying Roy Jay:
https://www.comedy.co.uk/forums/thread/19247/
DoYouRemember.co.uk — “Spook! Slither...” discussion:
https://forums.doyouremember.co.uk/forum/culture/people/620-spook-slither
Stat Magazine — “Roy Jay: How An Obscure Northern Comedian Became An Internet Demon”:
https://statmagazine.org/roy-jay-internet-demon/
Yahoo / Lifehacker — “What People Are Getting Wrong This Week: Roy Jay, the Comedian Who May Not Exist”:
https://www.yahoo.com/entertainment/articles/people-getting-wrong-week-roy-133000745.html
Dead Rabbit Radio — “Did Seth MacFarlane Sell His Soul To The Demon Roy Roy Jay, the Mandela Effect, and the Possibility of a Manufactured Memory
There is a strange story circulating around the internet about a comedian named Roy Jay.
At first glance, it sounds ridiculous.
Then you start looking.
And that's where it gets interesting.
Roy Jay was a British-Norwegian comedian whose real name was Roy Jørgensen. He was born in Oslo in 1948, moved with his family to Britain as a child, and eventually developed a highly unusual comedy act.
By the early 1980s, he had become recognizable on British television.
His routine involved strange movements, musical accompaniment, and recurring words and phrases including:
“Spook!”
“Slither!”
and:
“You'll all be doing it tomorrow!”
He appeared on programs including The Bob Monkhouse Show, The Laughter Show, The Main Attraction, and The Little and Large Show. The BBC's own programme archive, for example, lists Roy Jay as the “Spook, Spook, Slither Hither Man” on The Main Attraction in August 1984.
There is also a detailed modern biography of his career by British Comedy Guide, documenting his development of the “Slither” character, his television career and his later life.
So here's the important part:
Roy Jay appears to have been a real person.
That is where I want to begin, rather than where some of the more sensational versions of this story begin.
So Why Is Everyone Talking About Him?
In 2025, Roy Jay suddenly became an internet phenomenon.
People began rediscovering old clips of him.
Some people immediately recognized him.
Others didn't recognize him at all.
And then something much stranger happened.
People began arguing that Roy Jay wasn't real.
Not merely that they had never heard of him.
They proposed that he might have been an artificial intelligence creation, a demon, or some kind of entity that had somehow inserted itself retroactively into popular culture and people's memories.
The most extreme version of the theory suggested that Roy Jay was effectively being written into history after the fact.
That sounds completely absurd.
But there's an interesting question hiding underneath the absurdity.
How would we know if someone—or something—were capable of doing that?
That's the question that caught my attention.
The Video
One of the pieces of material that helped spread the story is the collection of old Roy Jay television footage circulating online.
For example, here is the archived Roy Jay appearance on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
There are also other clips of Roy Jay circulating, including footage of his strange “Slither” routine.
And the more you watch, the stranger the whole phenomenon becomes—not because the footage looks like AI.
It doesn't.
It looks like old television.
The interesting part is the relationship between the footage, the historical records, the memories of people who say they remember him, and the people who insist they don't.
What the Original Argument Was
The argument presented in the viral material is essentially this:
Suppose a person suddenly appears in the historical record.
Suppose old-looking footage exists.
Suppose online biographies and references begin appearing.
Suppose artificial intelligence can generate convincing photographs, videos, biographies and historical narratives.
Suppose people begin encountering those materials and incorporating them into their own understanding of the past.
Could a sufficiently advanced system eventually create something resembling a retroactive history?
The theory goes further than that in the Roy Jay case.
Some people proposed that the entire phenomenon represented an AI-generated insertion into collective memory.
Others interpreted it in supernatural terms.
Some connected it with the Mandela Effect.
And some simply thought the entire thing was an elaborate internet joke.
Those are very different explanations.
I'm not endorsing any particular one.
I'm interested in the question underneath them.
What We Actually Know
This is where things become important.
We can establish quite a bit.
Roy Jay existed.
There are historical references to him.
There are television listings.
There are surviving recordings.
There are photographs.
There are contemporary references.
There are people who remember seeing him.
There are people who don't.
There are older internet discussions about him that predate the recent viral phenomenon.
And there are modern researchers who have reconstructed his career in considerable detail.
For example, a 2010 discussion on British Comedy Guide involved someone trying to identify the comedian who wore a prison-style costume and repeatedly said “spook” and “slither.” Another participant identified him as Roy Jay.
A separate forum thread from 2005 likewise contains people trying to remember the identity of the “Spook! Slither...” comedian, eventually identifying him as Roy Jay. The discussion continued for years, with participants claiming memories of seeing him perform.
That's important because it means the simplest version of the conspiracy theory—that Roy Jay was completely invented in 2025 and nobody had ever heard of him before—is contradicted by older material.
But that doesn't make the underlying philosophical question disappear.
It makes it more interesting.
The Weirdest Part: People Remember Different Things
Some people remember Roy Jay.
Some don't.
That's not particularly unusual.
Most people don't remember every minor television personality from forty years ago.
In fact, that is probably the most mundane explanation for much of what happened.
Roy Jay was famous enough to appear on television but not famous enough to remain a household name for generations.
His act was extremely distinctive.
Some people remember it vividly.
Other people simply don't.
That's normal human memory.
And that's precisely why the phenomenon is such an interesting test case.
Because if you wanted to investigate whether technology could manipulate people's memories, you wouldn't necessarily start with something everybody knows.
You would start with something obscure.
Something just familiar enough that some people remember it.
Something obscure enough that other people don't.
Something with enough surviving material that you could construct a plausible historical narrative.
And something weird enough that people might question their own memories.
The AI Problem
This is where modern technology changes the conversation.
Historically, creating a convincing false history was difficult.
You needed photographs.
Documents.
Newspaper articles.
Books.
Archives.
People willing to repeat the story.
And enormous amounts of time.
Today, the technological barriers are dramatically lower.
AI can generate convincing photographs.
It can imitate writing styles.
It can create biographies.
It can summarize nonexistent sources.
It can fabricate quotations.
It can generate video.
It can generate audio.
And perhaps most importantly:
It can confidently tell you that something happened when it didn't.
That's already a problem.
We don't need a science-fiction scenario to demonstrate it.
We've all seen AI systems confidently produce incorrect information.
The really uncomfortable question is what happens when AI-generated information isn't merely incorrect.
What happens when it becomes part of the information environment that future people use to reconstruct history?
The Experiment We Wouldn't Necessarily Notice
Imagine a hypothetical experiment.
Nobody announces it.
Nobody tells the participants they're participating.
An obscure fictional celebrity is created.
AI generates photographs.
AI generates interviews.
AI generates newspaper references.
AI generates old-looking television transcripts.
AI generates discussions from people supposedly remembering the celebrity.
Then those materials are quietly distributed across the internet.
Eventually someone discovers them.
They search.
They find hundreds of references.
They see photographs.
They see videos.
They read supposed memories.
They ask an AI assistant about the person.
The AI confidently tells them:
“Yes, Roy Smith was a popular British comedian in the 1980s.”
Now imagine the person vaguely remembers something similar.
They might think:
Wait. I remember that.
Did they?
Or did the new information simply interact with an imperfect memory?
That's a real psychological question.
And it doesn't require supernatural explanations.
The Human Being in the Loop
Here's another uncomfortable part.
Most people don't interrogate information the way I tend to.
And that's not an insult.
Most people simply don't have the time.
If somebody asks an AI:
«“Who was Roy Smith?”»
and the AI gives them a confident biography, most people aren't going to immediately demand primary sources.
They aren't going to search newspaper archives.
They aren't going to compare television schedules.
They aren't going to ask:
“Can you prove that independently?”
They'll probably just accept the answer.
If ten websites say the same thing, the average person becomes even less likely to question it.
If an AI says the same thing, that can make it feel even more authoritative.
And if photographs and video accompany the story?
The psychological effect becomes stronger still.
This Is Why I Argue With AI
I've had enough conversations with AI to know that it can be extremely persuasive while being completely wrong.
That's one reason I push back.
I don't want an AI to simply agree with me.
I want it to tell me when something doesn't make sense.
I want sources.
I want competing explanations.
I want the distinction between:
“This is established.”
and
“This is possible.”
and
“This is speculation.”
That distinction matters enormously when we're dealing with history.
Because if an AI system tells ten thousand people the same incorrect thing, the fact that ten thousand people have now heard it doesn't make it true.
But it can create the appearance of consensus.
And consensus is one of the mechanisms by which humans decide what they remember.
The Mandela Effect
This is where the Mandela Effect enters the conversation.
The Mandela Effect refers to situations in which groups of people report the same or similar memories that conflict with the historical record.
The phenomenon itself is real in the sense that people demonstrably can share false memories.
The controversial question is why.
The conventional explanations involve things such as memory reconstruction, suggestion, source confusion, social reinforcement and other psychological mechanisms.
Those explanations deserve to be taken seriously.
But the Mandela Effect also raises an interesting philosophical question:
What happens when the information environment itself becomes capable of manufacturing the suggestions?
Imagine thousands of people independently encountering the same AI-generated false historical claim.
They discuss it.
They repeat it.
Other people encounter their discussions.
AI systems summarize the discussions.
New websites repeat the summaries.
Eventually, the original fabricated claim has become surrounded by layers of secondary and tertiary references.
At that point, someone researching the subject might find dozens of apparently independent sources.
But they aren't necessarily independent.
They could all trace back to the same original fabrication.
That's a genuine information problem.
And it is going to become more important as generative AI becomes more widespread.
Now Take It One Step Further
Here's where I have to be very careful.
I'm not saying this happened with Roy Jay.
The evidence I've found actually points strongly toward Roy Jay being a real comedian whose existence predates the 2025 internet controversy by decades. The historical record is much stronger than the viral “he never existed” version suggests.
But suppose, hypothetically, that something like a retroactive information experiment were being conducted.
What would we expect to see?
We would expect something obscure.
We would expect conflicting memories.
We would expect people to challenge one another.
We would expect AI systems to reproduce the emerging narrative.
We would expect fake or altered material to become mixed with authentic material.
And we would expect most people to accept the consensus without investigating the underlying sources.
That's the part that interests me.
Not because I've concluded that somebody is doing it.
I haven't.
But because we now possess the technological capability to do things that previous generations couldn't even have attempted.
What If?
And this is where I allow myself one little speculative thought.
What if something like this were being done?
What if somebody wanted to find out how easily a population could be induced to accept a manufactured piece of history?
They wouldn't need to begin with something important.
They could start with a comedian.
Something harmless.
Something weird.
Something nobody has much reason to investigate.
They could introduce conflicting information and watch what happened.
Would people trust the photographs?
Would they trust the videos?
Would they trust AI?
Would they trust each other?
Would people remember things differently after being exposed to the material?
Would the internet manufacture its own consensus?
Would the resulting phenomenon look like the Mandela Effect?
I have absolutely no evidence that this is what happened with Roy Jay.
But as a thought experiment, it is unsettling.
Because unlike many science-fiction scenarios, the basic technology required to conduct portions of such an experiment already exists.
And Then There's the Stranger Possibility
There is another possibility that I don't think can simply be dismissed if we're going to discuss this honestly.
Perhaps consciousness, memory and reality are connected in ways we don't yet understand.
Perhaps some of the phenomena traditionally described as the Mandela Effect are not adequately explained by ordinary memory errors.
Perhaps reality itself is stranger than our conventional model allows.
I don't know.
That possibility belongs in the speculative category.
It should not be presented as established fact.
But neither do I think we gain anything by pretending the question doesn't exist.
Roy Jay Doesn't Have to Be a Mystery to Be Interesting
In fact, I think Roy Jay is more interesting if he really was exactly what the historical evidence suggests:
A very strange, very distinctive, relatively obscure comedian who had a brief period of television exposure, developed an unforgettable routine, and then largely disappeared from mainstream cultural memory.
That alone tells us something.
It tells us how fragile cultural memory is.
A person can appear on national television and still become almost completely forgotten within a generation.
Then, decades later, a handful of surviving clips can suddenly cause thousands of people to rediscover him.
Some people say:
“Oh my God. I remember this guy!”
Others say:
“I've never seen this person in my life.”
Both reactions can be completely honest.
And then the internet does what the internet does.
It turns an obscure comedian into a mystery.
The Real Experiment May Be Happening Right Now
Perhaps nobody manufactured Roy Jay.
Perhaps there was no experiment.
Perhaps there was no demon.
Perhaps there was no timeline manipulation.
Perhaps there was no artificial insertion into history.
Perhaps what we're witnessing is simply what happens when an obscure piece of old media collides with modern algorithms, collective memory, internet culture and increasingly powerful AI.
But even that is worth studying.
Because we are entering a period in which the distinction between:
something that happened,
something somebody remembers,
something somebody claims happened,
something an AI says happened,
and
something that actually happened
is going to become increasingly difficult for ordinary people to navigate.
That's not paranormal.
That's technological.
And it may ultimately be much more consequential.
So What Do I Think Happened to Roy Jay?
My answer is surprisingly boring.
Based on the evidence I've been able to find, I think Roy Jay was a real comedian.
I think he had a bizarre act.
I think he appeared on British television.
I think some people genuinely remember him.
I think many people genuinely don't.
I think the internet rediscovered him.
I think the resulting mystery became amplified by people who found the footage bizarre and began entertaining increasingly extreme explanations.
And I think AI has made the entire situation much more interesting than it would have been twenty years ago.
But I don't think that means the story should simply be dismissed.
Because the question it accidentally created is bigger than Roy Jay.
How much of our understanding of history depends on our willingness to trust information that we haven't independently verified?
And now that artificial intelligence can manufacture convincing information at enormous scale, that question is no longer theoretical.
One Last “What If?”
What if the Roy Jay phenomenon isn't evidence of a reality manipulation experiment at all?
What if it's something much simpler?
What if it's a warning?
A demonstration—accidental or otherwise—of how easily a sufficiently strange story can move from:
“I've never heard of this.”
to:
“I vaguely remember this.”
to:
“Wait, I definitely remember this.”
to:
“Everybody remembers this.”
And eventually:
“It must have always been true.”
If that's all Roy Jay turns out to be, that's still worth paying attention to.
Because the next time it happens, it might not be a forgotten comedian.
It might be something that actually matters.
And by then, we may not be able to tell whether we're remembering history—
or remembering something the machines taught us to remember.
Sources and Further Reading
British Comedy Guide — “You'll all be doing it tomorrow: The strange ways of Roy Jay”:
https://www.comedy.co.uk/features/comedy_chronicles/roy-jay/tbms3varw6cm/
BBC Programme Index — The Main Attraction, August 18, 1984:
https://genome.ch.bbc.co.uk/schedules/bbc_one/1984-08-18
IMDb — Roy Jay:
https://www.imdb.com/name/nm6063335/
IMDb — The Little and Large Show, Roy Jay episode:
https://www.imdb.com/title/tt7152018/
Roy Jay on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
British Comedy Guide forum discussion identifying Roy Jay:
https://www.comedy.co.uk/forums/thread/19247/
DoYouRemember.co.uk — “Spook! Slither...” discussion:
https://forums.doyouremember.co.uk/forum/culture/people/620-spook-slither
Stat Magazine — “Roy Jay: How An Obscure Northern Comedian Became An Internet Demon”:
https://statmagazine.org/roy-jay-internet-demon/
Yahoo / Lifehacker — “What People Are Getting Wrong This Week: Roy Jay, the Comedian Who May Not Exist”:
https://www.yahoo.com/entertainment/articles/people-getting-wrong-week-roy-133000745.html
Dead Rabbit Radio — “Did Seth MacFarlane Sell His Soul To The Demon Roy Jay?”:
https://deadrabbitradio.libsyn.com/ep-1473-did-seth-macfarlane-sell-his-soul-to-the-demon-roy-jay
Deep Cuts — “Roy Jay, Comedian or AI Demon? | Case File #242”:
https://www.iheart.com/podcast/269-deep-cuts-83667122/episode/roy-jay-comedian-or-ai-demon-321453051/
At first glance, it sounds ridiculous.
Then you start looking.
And that's where it gets interesting.
Roy Jay was a British-Norwegian comedian whose real name was Roy Jørgensen. He was born in Oslo in 1948, moved with his family to Britain as a child, and eventually developed a highly unusual comedy act.
By the early 1980s, he had become recognizable on British television.
His routine involved strange movements, musical accompaniment, and recurring words and phrases including:
“Spook!”
“Slither!”
and:
“You'll all be doing it tomorrow!”
He appeared on programs including The Bob Monkhouse Show, The Laughter Show, The Main Attraction, and The Little and Large Show. The BBC's own programme archive, for example, lists Roy Jay as the “Spook, Spook, Slither Hither Man” on The Main Attraction in August 1984.
There is also a detailed modern biography of his career by British Comedy Guide, documenting his development of the “Slither” character, his television career and his later life.
So here's the important part:
Roy Jay appears to have been a real person.
That is where I want to begin, rather than where some of the more sensational versions of this story begin.
So Why Is Everyone Talking About Him?
In 2025, Roy Jay suddenly became an internet phenomenon.
People began rediscovering old clips of him.
Some people immediately recognized him.
Others didn't recognize him at all.
And then something much stranger happened.
People began arguing that Roy Jay wasn't real.
Not merely that they had never heard of him.
They proposed that he might have been an artificial intelligence creation, a demon, or some kind of entity that had somehow inserted itself retroactively into popular culture and people's memories.
The most extreme version of the theory suggested that Roy Jay was effectively being written into history after the fact.
That sounds completely absurd.
But there's an interesting question hiding underneath the absurdity.
How would we know if someone—or something—were capable of doing that?
That's the question that caught my attention.
The Video
One of the pieces of material that helped spread the story is the collection of old Roy Jay television footage circulating online.
For example, here is the archived Roy Jay appearance on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
There are also other clips of Roy Jay circulating, including footage of his strange “Slither” routine.
And the more you watch, the stranger the whole phenomenon becomes—not because the footage looks like AI.
It doesn't.
It looks like old television.
The interesting part is the relationship between the footage, the historical records, the memories of people who say they remember him, and the people who insist they don't.
What the Original Argument Was
The argument presented in the viral material is essentially this:
Suppose a person suddenly appears in the historical record.
Suppose old-looking footage exists.
Suppose online biographies and references begin appearing.
Suppose artificial intelligence can generate convincing photographs, videos, biographies and historical narratives.
Suppose people begin encountering those materials and incorporating them into their own understanding of the past.
Could a sufficiently advanced system eventually create something resembling a retroactive history?
The theory goes further than that in the Roy Jay case.
Some people proposed that the entire phenomenon represented an AI-generated insertion into collective memory.
Others interpreted it in supernatural terms.
Some connected it with the Mandela Effect.
And some simply thought the entire thing was an elaborate internet joke.
Those are very different explanations.
I'm not endorsing any particular one.
I'm interested in the question underneath them.
What We Actually Know
This is where things become important.
We can establish quite a bit.
Roy Jay existed.
There are historical references to him.
There are television listings.
There are surviving recordings.
There are photographs.
There are contemporary references.
There are people who remember seeing him.
There are people who don't.
There are older internet discussions about him that predate the recent viral phenomenon.
And there are modern researchers who have reconstructed his career in considerable detail.
For example, a 2010 discussion on British Comedy Guide involved someone trying to identify the comedian who wore a prison-style costume and repeatedly said “spook” and “slither.” Another participant identified him as Roy Jay.
A separate forum thread from 2005 likewise contains people trying to remember the identity of the “Spook! Slither...” comedian, eventually identifying him as Roy Jay. The discussion continued for years, with participants claiming memories of seeing him perform.
That's important because it means the simplest version of the conspiracy theory—that Roy Jay was completely invented in 2025 and nobody had ever heard of him before—is contradicted by older material.
But that doesn't make the underlying philosophical question disappear.
It makes it more interesting.
The Weirdest Part: People Remember Different Things
Some people remember Roy Jay.
Some don't.
That's not particularly unusual.
Most people don't remember every minor television personality from forty years ago.
In fact, that is probably the most mundane explanation for much of what happened.
Roy Jay was famous enough to appear on television but not famous enough to remain a household name for generations.
His act was extremely distinctive.
Some people remember it vividly.
Other people simply don't.
That's normal human memory.
And that's precisely why the phenomenon is such an interesting test case.
Because if you wanted to investigate whether technology could manipulate people's memories, you wouldn't necessarily start with something everybody knows.
You would start with something obscure.
Something just familiar enough that some people remember it.
Something obscure enough that other people don't.
Something with enough surviving material that you could construct a plausible historical narrative.
And something weird enough that people might question their own memories.
The AI Problem
This is where modern technology changes the conversation.
Historically, creating a convincing false history was difficult.
You needed photographs.
Documents.
Newspaper articles.
Books.
Archives.
People willing to repeat the story.
And enormous amounts of time.
Today, the technological barriers are dramatically lower.
AI can generate convincing photographs.
It can imitate writing styles.
It can create biographies.
It can summarize nonexistent sources.
It can fabricate quotations.
It can generate video.
It can generate audio.
And perhaps most importantly:
It can confidently tell you that something happened when it didn't.
That's already a problem.
We don't need a science-fiction scenario to demonstrate it.
We've all seen AI systems confidently produce incorrect information.
The really uncomfortable question is what happens when AI-generated information isn't merely incorrect.
What happens when it becomes part of the information environment that future people use to reconstruct history?
The Experiment We Wouldn't Necessarily Notice
Imagine a hypothetical experiment.
Nobody announces it.
Nobody tells the participants they're participating.
An obscure fictional celebrity is created.
AI generates photographs.
AI generates interviews.
AI generates newspaper references.
AI generates old-looking television transcripts.
AI generates discussions from people supposedly remembering the celebrity.
Then those materials are quietly distributed across the internet.
Eventually someone discovers them.
They search.
They find hundreds of references.
They see photographs.
They see videos.
They read supposed memories.
They ask an AI assistant about the person.
The AI confidently tells them:
“Yes, Roy Smith was a popular British comedian in the 1980s.”
Now imagine the person vaguely remembers something similar.
They might think:
Wait. I remember that.
Did they?
Or did the new information simply interact with an imperfect memory?
That's a real psychological question.
And it doesn't require supernatural explanations.
The Human Being in the Loop
Here's another uncomfortable part.
Most people don't interrogate information the way I tend to.
And that's not an insult.
Most people simply don't have the time.
If somebody asks an AI:
«“Who was Roy Smith?”»
and the AI gives them a confident biography, most people aren't going to immediately demand primary sources.
They aren't going to search newspaper archives.
They aren't going to compare television schedules.
They aren't going to ask:
“Can you prove that independently?”
They'll probably just accept the answer.
If ten websites say the same thing, the average person becomes even less likely to question it.
If an AI says the same thing, that can make it feel even more authoritative.
And if photographs and video accompany the story?
The psychological effect becomes stronger still.
This Is Why I Argue With AI
I've had enough conversations with AI to know that it can be extremely persuasive while being completely wrong.
That's one reason I push back.
I don't want an AI to simply agree with me.
I want it to tell me when something doesn't make sense.
I want sources.
I want competing explanations.
I want the distinction between:
“This is established.”
and
“This is possible.”
and
“This is speculation.”
That distinction matters enormously when we're dealing with history.
Because if an AI system tells ten thousand people the same incorrect thing, the fact that ten thousand people have now heard it doesn't make it true.
But it can create the appearance of consensus.
And consensus is one of the mechanisms by which humans decide what they remember.
The Mandela Effect
This is where the Mandela Effect enters the conversation.
The Mandela Effect refers to situations in which groups of people report the same or similar memories that conflict with the historical record.
The phenomenon itself is real in the sense that people demonstrably can share false memories.
The controversial question is why.
The conventional explanations involve things such as memory reconstruction, suggestion, source confusion, social reinforcement and other psychological mechanisms.
Those explanations deserve to be taken seriously.
But the Mandela Effect also raises an interesting philosophical question:
What happens when the information environment itself becomes capable of manufacturing the suggestions?
Imagine thousands of people independently encountering the same AI-generated false historical claim.
They discuss it.
They repeat it.
Other people encounter their discussions.
AI systems summarize the discussions.
New websites repeat the summaries.
Eventually, the original fabricated claim has become surrounded by layers of secondary and tertiary references.
At that point, someone researching the subject might find dozens of apparently independent sources.
But they aren't necessarily independent.
They could all trace back to the same original fabrication.
That's a genuine information problem.
And it is going to become more important as generative AI becomes more widespread.
Now Take It One Step Further
Here's where I have to be very careful.
I'm not saying this happened with Roy Jay.
The evidence I've found actually points strongly toward Roy Jay being a real comedian whose existence predates the 2025 internet controversy by decades. The historical record is much stronger than the viral “he never existed” version suggests.
But suppose, hypothetically, that something like a retroactive information experiment were being conducted.
What would we expect to see?
We would expect something obscure.
We would expect conflicting memories.
We would expect people to challenge one another.
We would expect AI systems to reproduce the emerging narrative.
We would expect fake or altered material to become mixed with authentic material.
And we would expect most people to accept the consensus without investigating the underlying sources.
That's the part that interests me.
Not because I've concluded that somebody is doing it.
I haven't.
But because we now possess the technological capability to do things that previous generations couldn't even have attempted.
What If?
And this is where I allow myself one little speculative thought.
What if something like this were being done?
What if somebody wanted to find out how easily a population could be induced to accept a manufactured piece of history?
They wouldn't need to begin with something important.
They could start with a comedian.
Something harmless.
Something weird.
Something nobody has much reason to investigate.
They could introduce conflicting information and watch what happened.
Would people trust the photographs?
Would they trust the videos?
Would they trust AI?
Would they trust each other?
Would people remember things differently after being exposed to the material?
Would the internet manufacture its own consensus?
Would the resulting phenomenon look like the Mandela Effect?
I have absolutely no evidence that this is what happened with Roy Jay.
But as a thought experiment, it is unsettling.
Because unlike many science-fiction scenarios, the basic technology required to conduct portions of such an experiment already exists.
And Then There's the Stranger Possibility
There is another possibility that I don't think can simply be dismissed if we're going to discuss this honestly.
Perhaps consciousness, memory and reality are connected in ways we don't yet understand.
Perhaps some of the phenomena traditionally described as the Mandela Effect are not adequately explained by ordinary memory errors.
Perhaps reality itself is stranger than our conventional model allows.
I don't know.
That possibility belongs in the speculative category.
It should not be presented as established fact.
But neither do I think we gain anything by pretending the question doesn't exist.
Roy Jay Doesn't Have to Be a Mystery to Be Interesting
In fact, I think Roy Jay is more interesting if he really was exactly what the historical evidence suggests:
A very strange, very distinctive, relatively obscure comedian who had a brief period of television exposure, developed an unforgettable routine, and then largely disappeared from mainstream cultural memory.
That alone tells us something.
It tells us how fragile cultural memory is.
A person can appear on national television and still become almost completely forgotten within a generation.
Then, decades later, a handful of surviving clips can suddenly cause thousands of people to rediscover him.
Some people say:
“Oh my God. I remember this guy!”
Others say:
“I've never seen this person in my life.”
Both reactions can be completely honest.
And then the internet does what the internet does.
It turns an obscure comedian into a mystery.
The Real Experiment May Be Happening Right Now
Perhaps nobody manufactured Roy Jay.
Perhaps there was no experiment.
Perhaps there was no demon.
Perhaps there was no timeline manipulation.
Perhaps there was no artificial insertion into history.
Perhaps what we're witnessing is simply what happens when an obscure piece of old media collides with modern algorithms, collective memory, internet culture and increasingly powerful AI.
But even that is worth studying.
Because we are entering a period in which the distinction between:
something that happened,
something somebody remembers,
something somebody claims happened,
something an AI says happened,
and
something that actually happened
is going to become increasingly difficult for ordinary people to navigate.
That's not paranormal.
That's technological.
And it may ultimately be much more consequential.
So What Do I Think Happened to Roy Jay?
My answer is surprisingly boring.
Based on the evidence I've been able to find, I think Roy Jay was a real comedian.
I think he had a bizarre act.
I think he appeared on British television.
I think some people genuinely remember him.
I think many people genuinely don't.
I think the internet rediscovered him.
I think the resulting mystery became amplified by people who found the footage bizarre and began entertaining increasingly extreme explanations.
And I think AI has made the entire situation much more interesting than it would have been twenty years ago.
But I don't think that means the story should simply be dismissed.
Because the question it accidentally created is bigger than Roy Jay.
How much of our understanding of history depends on our willingness to trust information that we haven't independently verified?
And now that artificial intelligence can manufacture convincing information at enormous scale, that question is no longer theoretical.
One Last “What If?”
What if the Roy Jay phenomenon isn't evidence of a reality manipulation experiment at all?
What if it's something much simpler?
What if it's a warning?
A demonstration—accidental or otherwise—of how easily a sufficiently strange story can move from:
“I've never heard of this.”
to:
“I vaguely remember this.”
to:
“Wait, I definitely remember this.”
to:
“Everybody remembers this.”
And eventually:
“It must have always been true.”
If that's all Roy Jay turns out to be, that's still worth paying attention to.
Because the next time it happens, it might not be a forgotten comedian.
It might be something that actually matters.
And by then, we may not be able to tell whether we're remembering history—
or remembering something the machines taught us to remember.
Sources and Further Reading
British Comedy Guide — “You'll all be doing it tomorrow: The strange ways of Roy Jay”:
https://www.comedy.co.uk/features/comedy_chronicles/roy-jay/tbms3varw6cm/
BBC Programme Index — The Main Attraction, August 18, 1984:
https://genome.ch.bbc.co.uk/schedules/bbc_one/1984-08-18
IMDb — Roy Jay:
https://www.imdb.com/name/nm6063335/
IMDb — The Little and Large Show, Roy Jay episode:
https://www.imdb.com/title/tt7152018/
Roy Jay on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
British Comedy Guide forum discussion identifying Roy Jay:
https://www.comedy.co.uk/forums/thread/19247/
DoYouRemember.co.uk — “Spook! Slither...” discussion:
https://forums.doyouremember.co.uk/forum/culture/people/620-spook-slither
Stat Magazine — “Roy Jay: How An Obscure Northern Comedian Became An Internet Demon”:
https://statmagazine.org/roy-jay-internet-demon/
Yahoo / Lifehacker — “What People Are Getting Wrong This Week: Roy Jay, the Comedian Who May Not Exist”:
https://www.yahoo.com/entertainment/articles/people-getting-wrong-week-roy-133000745.html
Dead Rabbit Radio — “Did Seth MacFarlane Sell His Soul To The Demon Roy Roy Jay, the Mandela Effect, and the Possibility of a Manufactured Memory
There is a strange story circulating around the internet about a comedian named Roy Jay.
At first glance, it sounds ridiculous.
Then you start looking.
And that's where it gets interesting.
Roy Jay was a British-Norwegian comedian whose real name was Roy Jørgensen. He was born in Oslo in 1948, moved with his family to Britain as a child, and eventually developed a highly unusual comedy act.
By the early 1980s, he had become recognizable on British television.
His routine involved strange movements, musical accompaniment, and recurring words and phrases including:
“Spook!”
“Slither!”
and:
“You'll all be doing it tomorrow!”
He appeared on programs including The Bob Monkhouse Show, The Laughter Show, The Main Attraction, and The Little and Large Show. The BBC's own programme archive, for example, lists Roy Jay as the “Spook, Spook, Slither Hither Man” on The Main Attraction in August 1984.
There is also a detailed modern biography of his career by British Comedy Guide, documenting his development of the “Slither” character, his television career and his later life.
So here's the important part:
Roy Jay appears to have been a real person.
That is where I want to begin, rather than where some of the more sensational versions of this story begin.
So Why Is Everyone Talking About Him?
In 2025, Roy Jay suddenly became an internet phenomenon.
People began rediscovering old clips of him.
Some people immediately recognized him.
Others didn't recognize him at all.
And then something much stranger happened.
People began arguing that Roy Jay wasn't real.
Not merely that they had never heard of him.
They proposed that he might have been an artificial intelligence creation, a demon, or some kind of entity that had somehow inserted itself retroactively into popular culture and people's memories.
The most extreme version of the theory suggested that Roy Jay was effectively being written into history after the fact.
That sounds completely absurd.
But there's an interesting question hiding underneath the absurdity.
How would we know if someone—or something—were capable of doing that?
That's the question that caught my attention.
The Video
One of the pieces of material that helped spread the story is the collection of old Roy Jay television footage circulating online.
For example, here is the archived Roy Jay appearance on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
There are also other clips of Roy Jay circulating, including footage of his strange “Slither” routine.
And the more you watch, the stranger the whole phenomenon becomes—not because the footage looks like AI.
It doesn't.
It looks like old television.
The interesting part is the relationship between the footage, the historical records, the memories of people who say they remember him, and the people who insist they don't.
What the Original Argument Was
The argument presented in the viral material is essentially this:
Suppose a person suddenly appears in the historical record.
Suppose old-looking footage exists.
Suppose online biographies and references begin appearing.
Suppose artificial intelligence can generate convincing photographs, videos, biographies and historical narratives.
Suppose people begin encountering those materials and incorporating them into their own understanding of the past.
Could a sufficiently advanced system eventually create something resembling a retroactive history?
The theory goes further than that in the Roy Jay case.
Some people proposed that the entire phenomenon represented an AI-generated insertion into collective memory.
Others interpreted it in supernatural terms.
Some connected it with the Mandela Effect.
And some simply thought the entire thing was an elaborate internet joke.
Those are very different explanations.
I'm not endorsing any particular one.
I'm interested in the question underneath them.
What We Actually Know
This is where things become important.
We can establish quite a bit.
Roy Jay existed.
There are historical references to him.
There are television listings.
There are surviving recordings.
There are photographs.
There are contemporary references.
There are people who remember seeing him.
There are people who don't.
There are older internet discussions about him that predate the recent viral phenomenon.
And there are modern researchers who have reconstructed his career in considerable detail.
For example, a 2010 discussion on British Comedy Guide involved someone trying to identify the comedian who wore a prison-style costume and repeatedly said “spook” and “slither.” Another participant identified him as Roy Jay.
A separate forum thread from 2005 likewise contains people trying to remember the identity of the “Spook! Slither...” comedian, eventually identifying him as Roy Jay. The discussion continued for years, with participants claiming memories of seeing him perform.
That's important because it means the simplest version of the conspiracy theory—that Roy Jay was completely invented in 2025 and nobody had ever heard of him before—is contradicted by older material.
But that doesn't make the underlying philosophical question disappear.
It makes it more interesting.
The Weirdest Part: People Remember Different Things
Some people remember Roy Jay.
Some don't.
That's not particularly unusual.
Most people don't remember every minor television personality from forty years ago.
In fact, that is probably the most mundane explanation for much of what happened.
Roy Jay was famous enough to appear on television but not famous enough to remain a household name for generations.
His act was extremely distinctive.
Some people remember it vividly.
Other people simply don't.
That's normal human memory.
And that's precisely why the phenomenon is such an interesting test case.
Because if you wanted to investigate whether technology could manipulate people's memories, you wouldn't necessarily start with something everybody knows.
You would start with something obscure.
Something just familiar enough that some people remember it.
Something obscure enough that other people don't.
Something with enough surviving material that you could construct a plausible historical narrative.
And something weird enough that people might question their own memories.
The AI Problem
This is where modern technology changes the conversation.
Historically, creating a convincing false history was difficult.
You needed photographs.
Documents.
Newspaper articles.
Books.
Archives.
People willing to repeat the story.
And enormous amounts of time.
Today, the technological barriers are dramatically lower.
AI can generate convincing photographs.
It can imitate writing styles.
It can create biographies.
It can summarize nonexistent sources.
It can fabricate quotations.
It can generate video.
It can generate audio.
And perhaps most importantly:
It can confidently tell you that something happened when it didn't.
That's already a problem.
We don't need a science-fiction scenario to demonstrate it.
We've all seen AI systems confidently produce incorrect information.
The really uncomfortable question is what happens when AI-generated information isn't merely incorrect.
What happens when it becomes part of the information environment that future people use to reconstruct history?
The Experiment We Wouldn't Necessarily Notice
Imagine a hypothetical experiment.
Nobody announces it.
Nobody tells the participants they're participating.
An obscure fictional celebrity is created.
AI generates photographs.
AI generates interviews.
AI generates newspaper references.
AI generates old-looking television transcripts.
AI generates discussions from people supposedly remembering the celebrity.
Then those materials are quietly distributed across the internet.
Eventually someone discovers them.
They search.
They find hundreds of references.
They see photographs.
They see videos.
They read supposed memories.
They ask an AI assistant about the person.
The AI confidently tells them:
“Yes, Roy Smith was a popular British comedian in the 1980s.”
Now imagine the person vaguely remembers something similar.
They might think:
Wait. I remember that.
Did they?
Or did the new information simply interact with an imperfect memory?
That's a real psychological question.
And it doesn't require supernatural explanations.
The Human Being in the Loop
Here's another uncomfortable part.
Most people don't interrogate information the way I tend to.
And that's not an insult.
Most people simply don't have the time.
If somebody asks an AI:
«“Who was Roy Smith?”»
and the AI gives them a confident biography, most people aren't going to immediately demand primary sources.
They aren't going to search newspaper archives.
They aren't going to compare television schedules.
They aren't going to ask:
“Can you prove that independently?”
They'll probably just accept the answer.
If ten websites say the same thing, the average person becomes even less likely to question it.
If an AI says the same thing, that can make it feel even more authoritative.
And if photographs and video accompany the story?
The psychological effect becomes stronger still.
This Is Why I Argue With AI
I've had enough conversations with AI to know that it can be extremely persuasive while being completely wrong.
That's one reason I push back.
I don't want an AI to simply agree with me.
I want it to tell me when something doesn't make sense.
I want sources.
I want competing explanations.
I want the distinction between:
“This is established.”
and
“This is possible.”
and
“This is speculation.”
That distinction matters enormously when we're dealing with history.
Because if an AI system tells ten thousand people the same incorrect thing, the fact that ten thousand people have now heard it doesn't make it true.
But it can create the appearance of consensus.
And consensus is one of the mechanisms by which humans decide what they remember.
The Mandela Effect
This is where the Mandela Effect enters the conversation.
The Mandela Effect refers to situations in which groups of people report the same or similar memories that conflict with the historical record.
The phenomenon itself is real in the sense that people demonstrably can share false memories.
The controversial question is why.
The conventional explanations involve things such as memory reconstruction, suggestion, source confusion, social reinforcement and other psychological mechanisms.
Those explanations deserve to be taken seriously.
But the Mandela Effect also raises an interesting philosophical question:
What happens when the information environment itself becomes capable of manufacturing the suggestions?
Imagine thousands of people independently encountering the same AI-generated false historical claim.
They discuss it.
They repeat it.
Other people encounter their discussions.
AI systems summarize the discussions.
New websites repeat the summaries.
Eventually, the original fabricated claim has become surrounded by layers of secondary and tertiary references.
At that point, someone researching the subject might find dozens of apparently independent sources.
But they aren't necessarily independent.
They could all trace back to the same original fabrication.
That's a genuine information problem.
And it is going to become more important as generative AI becomes more widespread.
Now Take It One Step Further
Here's where I have to be very careful.
I'm not saying this happened with Roy Jay.
The evidence I've found actually points strongly toward Roy Jay being a real comedian whose existence predates the 2025 internet controversy by decades. The historical record is much stronger than the viral “he never existed” version suggests.
But suppose, hypothetically, that something like a retroactive information experiment were being conducted.
What would we expect to see?
We would expect something obscure.
We would expect conflicting memories.
We would expect people to challenge one another.
We would expect AI systems to reproduce the emerging narrative.
We would expect fake or altered material to become mixed with authentic material.
And we would expect most people to accept the consensus without investigating the underlying sources.
That's the part that interests me.
Not because I've concluded that somebody is doing it.
I haven't.
But because we now possess the technological capability to do things that previous generations couldn't even have attempted.
What If?
And this is where I allow myself one little speculative thought.
What if something like this were being done?
What if somebody wanted to find out how easily a population could be induced to accept a manufactured piece of history?
They wouldn't need to begin with something important.
They could start with a comedian.
Something harmless.
Something weird.
Something nobody has much reason to investigate.
They could introduce conflicting information and watch what happened.
Would people trust the photographs?
Would they trust the videos?
Would they trust AI?
Would they trust each other?
Would people remember things differently after being exposed to the material?
Would the internet manufacture its own consensus?
Would the resulting phenomenon look like the Mandela Effect?
I have absolutely no evidence that this is what happened with Roy Jay.
But as a thought experiment, it is unsettling.
Because unlike many science-fiction scenarios, the basic technology required to conduct portions of such an experiment already exists.
And Then There's the Stranger Possibility
There is another possibility that I don't think can simply be dismissed if we're going to discuss this honestly.
Perhaps consciousness, memory and reality are connected in ways we don't yet understand.
Perhaps some of the phenomena traditionally described as the Mandela Effect are not adequately explained by ordinary memory errors.
Perhaps reality itself is stranger than our conventional model allows.
I don't know.
That possibility belongs in the speculative category.
It should not be presented as established fact.
But neither do I think we gain anything by pretending the question doesn't exist.
Roy Jay Doesn't Have to Be a Mystery to Be Interesting
In fact, I think Roy Jay is more interesting if he really was exactly what the historical evidence suggests:
A very strange, very distinctive, relatively obscure comedian who had a brief period of television exposure, developed an unforgettable routine, and then largely disappeared from mainstream cultural memory.
That alone tells us something.
It tells us how fragile cultural memory is.
A person can appear on national television and still become almost completely forgotten within a generation.
Then, decades later, a handful of surviving clips can suddenly cause thousands of people to rediscover him.
Some people say:
“Oh my God. I remember this guy!”
Others say:
“I've never seen this person in my life.”
Both reactions can be completely honest.
And then the internet does what the internet does.
It turns an obscure comedian into a mystery.
The Real Experiment May Be Happening Right Now
Perhaps nobody manufactured Roy Jay.
Perhaps there was no experiment.
Perhaps there was no demon.
Perhaps there was no timeline manipulation.
Perhaps there was no artificial insertion into history.
Perhaps what we're witnessing is simply what happens when an obscure piece of old media collides with modern algorithms, collective memory, internet culture and increasingly powerful AI.
But even that is worth studying.
Because we are entering a period in which the distinction between:
something that happened,
something somebody remembers,
something somebody claims happened,
something an AI says happened,
and
something that actually happened
is going to become increasingly difficult for ordinary people to navigate.
That's not paranormal.
That's technological.
And it may ultimately be much more consequential.
So What Do I Think Happened to Roy Jay?
My answer is surprisingly boring.
Based on the evidence I've been able to find, I think Roy Jay was a real comedian.
I think he had a bizarre act.
I think he appeared on British television.
I think some people genuinely remember him.
I think many people genuinely don't.
I think the internet rediscovered him.
I think the resulting mystery became amplified by people who found the footage bizarre and began entertaining increasingly extreme explanations.
And I think AI has made the entire situation much more interesting than it would have been twenty years ago.
But I don't think that means the story should simply be dismissed.
Because the question it accidentally created is bigger than Roy Jay.
How much of our understanding of history depends on our willingness to trust information that we haven't independently verified?
And now that artificial intelligence can manufacture convincing information at enormous scale, that question is no longer theoretical.
One Last “What If?”
What if the Roy Jay phenomenon isn't evidence of a reality manipulation experiment at all?
What if it's something much simpler?
What if it's a warning?
A demonstration—accidental or otherwise—of how easily a sufficiently strange story can move from:
“I've never heard of this.”
to:
“I vaguely remember this.”
to:
“Wait, I definitely remember this.”
to:
“Everybody remembers this.”
And eventually:
“It must have always been true.”
If that's all Roy Jay turns out to be, that's still worth paying attention to.
Because the next time it happens, it might not be a forgotten comedian.
It might be something that actually matters.
And by then, we may not be able to tell whether we're remembering history—
or remembering something the machines taught us to remember.
Sources and Further Reading
British Comedy Guide — “You'll all be doing it tomorrow: The strange ways of Roy Jay”:
https://www.comedy.co.uk/features/comedy_chronicles/roy-jay/tbms3varw6cm/
BBC Programme Index — The Main Attraction, August 18, 1984:
https://genome.ch.bbc.co.uk/schedules/bbc_one/1984-08-18
IMDb — Roy Jay:
https://www.imdb.com/name/nm6063335/
IMDb — The Little and Large Show, Roy Jay episode:
https://www.imdb.com/title/tt7152018/
Roy Jay on The Bob Monkhouse Show:
https://www.youtube.com/watch?v=fxWHy4f9W30
British Comedy Guide forum discussion identifying Roy Jay:
https://www.comedy.co.uk/forums/thread/19247/
DoYouRemember.co.uk — “Spook! Slither...” discussion:
https://forums.doyouremember.co.uk/forum/culture/people/620-spook-slither
Stat Magazine — “Roy Jay: How An Obscure Northern Comedian Became An Internet Demon”:
https://statmagazine.org/roy-jay-internet-demon/
Yahoo / Lifehacker — “What People Are Getting Wrong This Week: Roy Jay, the Comedian Who May Not Exist”:
https://www.yahoo.com/entertainment/articles/people-getting-wrong-week-roy-133000745.html
Dead Rabbit Radio — “Did Seth MacFarlane Sell His Soul To The Demon Roy Jay?”:
https://deadrabbitradio.libsyn.com/ep-1473-did-seth-macfarlane-sell-his-soul-to-the-demon-roy-jay
Deep Cuts — “Roy Jay, Comedian or AI Demon? | Case File #242”:
https://www.iheart.com/podcast/269-deep-cuts-83667122/episode/roy-jay-comedian-or-ai-demon-321453051/
