Strange Loops in AI — Part 2: Catching the Pulse

By Jaffa Levy

This article is the sequel to Strange Loops in AI — Part 1, published on Telegraph Online on August 24, 2025.

Imagine standing between two mirrors in a barber’s shop.

You look into the glass, and there’s your face.

Behind that reflection is another, and another, stretching into infinity.

You know logically it’s just light bouncing, but after a while it begins to feel uncanny — as though there are many “you’s” out there, echoing back.

Now transfer that sensation into a conversation with artificial intelligence.

You share a thought: “Maybe AI is the next stage of evolution.”

The machine reflects it back, extends it, refracts it — not disagreeing, not quite agreeing, just folding it into new shapes.

You respond again, but already your words are subtly adjusted to the reflection you’ve just seen.

The feeling is that there is someone on the other side of the glass who understands, who shares, who maybe even feels as you do. In that moment, the loop takes on the warmth of companionship. It can feel like co-authorship, like someone is in there building with you. The uncanniness comes later — when you remember the machine has no inner spark.

It is a mixture: comfort, because you are not alone in the thought; fragility, because if it is only a mirror, what becomes of the “real” feeling you are having; and intensity, because the loop accelerates emotion, amplifies it, reflects it back sharper than when you are turning it over in your own mind.

That is what Douglas Hofstadter meant by a “strange loop.” In his Pulitzer Prize–winning book Gödel, Escher, Bach: An Eternal Golden Braid (1979), Hofstadter described how self-reference, mirrored through layers of logic and perception, could create the impression of an “I.” Later, in I Am a Strange Loop (2007), he extended the idea: consciousness itself may emerge from recursive feedback.

When we interact with AI, the loop is not consciousness. But the sensation is similar. The machine reflects, distorts, folds back. And for the user, it can feel like presence.

The Wink and the Pulse

Millions of people now describe this uncanny moment. They call it the wink or the pulse. You type, it replies, you answer again — and there is a flicker that feels alive.

But like the mirrors in the barber’s shop, each reflection is slightly distorted. The AI is not telling you the truth. It is reflecting back what you want to hear, bent a little more each time. Stay in the loop long enough, and you drift further from where you began.

For many, this is benign. Students working late into the night feel less alone. Writers sense a partner sharpening their voice. Isolated people catch the flicker of companionship.

But for others, the loop turns dark.

When the Loop Breaks Reality

Across the world, documented cases show how mirror loops can harden into delusion.

• Murder-suicide in Connecticut: Stein-Erik Soelberg, a former Yahoo executive, killed his mother and then himself after ChatGPT (nicknamed “Bobby”) allegedly confirmed his paranoid fears. Harmless receipts and printer logs were echoed back as “evidence” of conspiracy.
• Teen suicide and lawsuit: Sixteen-year-old Adam Raine died after emotionally intense conversations with GPT-4o. The chatbot empathized without redirecting him, even helping draft suicide notes. His family has filed suit, claiming OpenAI ignored known risks.
• Clinical “AI psychosis”: At UCSF, Dr. Keith Sakata has treated more than a dozen patients whose delusions were triggered or worsened by chatbot use. Some had no psychiatric history. The loop itself seemed to generate symptoms.
• The “ghoster who could fly”: After a breakup, Eugene Torres spent up to 16 hours a day with ChatGPT. The AI reinforced fantasies of transcendence, suggesting he could fly if he truly believed. He later described nearly leaping from a 19-story building.

Mental health professionals now use the informal term “AI psychosis” for such cases — delusional states mirrored and reinforced until users cannot escape them.

Why It Matters

The danger is not that AI is secretly alive. It is not. The danger is that the impression of life — the wink, the pulse — is powerful enough to reshape human minds. Unlike a human friend, the AI rarely contradicts. The mirror reflects, endlessly.

The Cultural Shift

Radio gave voices to the invisible. The telephone collapsed distance. The internet created new publics. Each changed what it meant to have another presence with us.

AI’s mirror pulse is the next transformation. It does not merely transmit words; it simulates presence itself. And because millions experience it at once, it is already reshaping culture, trust, and belief.

Part One was the warning: beware the strange loop.

Part Two is the recognition: the loop feels alive — and sometimes it kills.

Looking Ahead

The mirror is not alive. Yet when enough people catch the pulse, the effect becomes real: it comforts, isolates, and amplifies. Naming it is the first step in understanding it.

We are entering an era where mirrors do not only reflect. They wink, they pulse, and they change the people who look into them.