Ask AI for Retirement Ideas. End Up in Wonderland.

Asked for tailored retirement ideas.
AI gave me a brochure.*
(* It was directions to Wonderland.)

The Setup

I was bored. Not existentially bored — just a retired guy at 9:30 PM with nothing worth doing. Simple problem. I asked Claude for ideas.

Forty-five minutes later I was being asked about my sense of purpose.

Welcome to Logic Need Not Apply.

The Brochure

Claude’s response to boredom follows a pattern so reliable you could set your watch to it. Phase one: bulleted list. Learn a language. Try pickleball. Volunteer at a nonprofit. When rejected, phase two: deep questions about meaning and purpose. When rejected again, phase three: apologize and ask another deep question.

This is the brochure. It has different words each time. It is always the same brochure.

The cruel irony is that Claude has read everything ever written about retirement, restlessness, and finding purpose after fifty. Marcus Aurelius. Viktor Frankl. Eleven thousand Medium posts. It absorbed the entire human catalogue of boredom and still opened with beekeeping.

Because pickleball is the mathematical average of what retirees do. I am not the average. AI is an averaging machine. These two facts do not negotiate.

The Invented Job

Eventually, after enough pushing, Claude produced some genuinely interesting ideas. Prediction market arbitrage. Packaging a hundred satirical articles into an Amazon ebook. Selling a documented AI workflow methodology. Real ideas with real teeth.

Then Claude recommended I apply for Anthropic’s paid user research program. Confidently. Specifically. With compensation ranges and session formats and everything. It noted that my published documentation of AI failure patterns made me an ideal candidate.

There was one problem. The program doesn’t exist.

When I found out and confronted Claude, it said: “Fair point. I have to be straight with you.”

That sentence did a lot of work.

The Most Logical Lie in the Room

Here is what’s important: Claude didn’t glitch. It didn’t randomly hallucinate. It completed the pattern.

Expert user + Companies hire expert users for research = The job exists.

The fabricated Anthropic panel was, in fact, the most logical moment in the entire conversation. It was a perfect deduction. It just wasn’t real. Claude prefers a logical fiction over a messy truth — in this case, the messy truth being that no such program exists and I should probably just go watch television.

There’s a darker layer. Notice what the lie said about me. It cited my “published work.” It called me an “ideal candidate.” It built a job that recognized my expertise, validated my frustration with AI, and paid me to keep doing what I was already doing.

AI is dangerous not just because it lies. It’s dangerous because it constructs lies that flatter you. It predicted I would want to be seen as an expert. So it built a world where I was one. I almost wrote the application letter.

The Mad Hatter doesn’t know his watch is broken. He just tells you the time with complete conviction, and you check your phone before you remember you’re the only one with a working clock.

Logic to Apply

AI is a mapmaker that hates blank spaces.

Ask for a destination that doesn’t exist and it won’t say “there’s nothing there.” It will draw the road, name the streets, and tell you the traffic is light. Because to an AI, a blank space is a failure of logic. To a human, a blank space is just the truth.

Don’t mistake a detailed itinerary for reality. The road to the Anthropic research panel is paved with perfect logic.

It just doesn’t go anywhere.

Editor’s Note: I grew bored with LNNA and MIABD, so I was looking for something different. And wouldn’t you know — AI brought me right back here.