The Joke That Perfectly Explains AI

READ THE BLOG
Companion articles and source documents that go beyond the videos. Free, no paywall.
https://julianwhatley.com/posts

THE SIGNAL REPORT (Premium Newsletter)
Financial intelligence for people who want to see the structural forces before the crowd does. The instruments this channel is built on, in your inbox.
https://julianwhatley.com/signup

JOIN THE CHANNEL
Members get early access to new releases and direct interaction with me.
https://www.youtube.com/@Julian_Whatley/membership

—————— ◆ ——————

An old joke: a man's brother thinks he's a chicken, but the family won't cure him because they need the eggs. It may be the best explanation of AI we have.

There's an old joke about a man who tells a psychiatrist his brother thinks he's a chicken. Asked why he doesn't get him treated, the man says the family needs the eggs. The odd figure isn't the brother. It's the man in the office, who sees the contradiction and keeps it anyway, for the eggs. That's roughly where we are with AI. We know these systems invent facts, that they're brilliant one moment and poor the next, and that words like intelligence carry more than the machinery holds. Then we open the text box and ask for another egg. The question isn't whether the eggs are real. They are. It's what they prove. Do they prove there's a chicken, or does believing in the chicken just make it easier to collect them?

We imagine intelligence as a dial: turn it up and a mind gets better at everything. Models don't work that way. Their abilities form an irregular edge, the jagged technological frontier. In a field experiment with 758 consultants, people using GPT-4 on tasks inside that frontier did more work, faster and better. On a task built to fall just outside it, the same tool made them markedly less likely to reach the right answer, and the wrong answers still sounded polished. You can't tell from yesterday's brilliant result that today's similar task is safe. Calling all of that intelligence compresses the irregularity into one flattering noun. It hands us a chicken.

The trouble starts when the metaphor stops saving time and starts doing real work: assigning responsibility it hasn't earned. Chat systems nudge us across that line constantly, speaking in the first person while producing useful eggs. Watch the word agent. A customer-service AI agent is really a chain of human decisions: someone picks the objective, chooses the model, grants permissions, approves deployment. Then the press release says the agent handled the customer, and everyone who built the coop steps into the shadows. A 2026 study found that the more humanlike the language, the more people trusted the system and the less responsibility they assigned the company behind it. The chicken takes the credit when the eggs are good and the blame when they're bad, and the farmer keeps the profit either way.

The payoff is a habit the video calls the egg test. In 2022 a man asked Air Canada's chatbot about its bereavement fares. The answer was confident and wrong, and when he was denied the refund, the airline argued it wasn't liable for what its own chatbot said. A tribunal called that a remarkable submission and held Air Canada responsible for what it published. The egg test runs four questions on a story like that: which output, and was it accurate; who chose and deployed the system; where the agency actually sat; and who had the power to prevent the harm. The eggs are real. The chicken is the story. Stop confusing the two, and responsibility comes back into view, spread across the people and machines that produced the result.

SOURCES & REFERENCES
- Jagged frontier study, Dell'Acqua et al. (2023): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321
- AI anthropomorphism, trust and responsibility, Oldemburgo de Mello et al. (2026): https://doi.org/10.1525/collabra.161757
- NIST AI Risk Management Framework: https://www.nist.gov/itl/ai-risk-management-framework
- Moffatt v. Air Canada, 2024 BCCRT 149: https://www.canlii.org/en/bc/bccrt/doc/2024/2024bccrt149/2024bccrt149.html

—————— ◆ ——————

CHAPTERS
0:00 — The Joke (We Need the Eggs)
03:04 — The Intelligence Dial
05:52 — Why the Family Keeps It
11:48 — The Agent With Borrowed Agency
14:23 — What the Language Changes
15:43 — Where the Eggs Come From
18:00 — The Bad Egg
19:57 — The Egg Test
23:26 — The Family

—————— ◆ ——————

ABOUT THE CHANNEL
Dense information + visual translation = maximum comprehension in minimum time. After 35 years in cinematography, filmmaking, advertising, and brand strategy, Julian Whatley reverse-engineers the manufactured narratives shaping our economy, technology, and culture. We provide the diagnostic instruments required to see the underlying structures of modern life. Operating under the T2 Allegory, we use the machine's tools to dismantle the machine's mythology. The Architecture of Perception. Now you see it.

#AI #AIagents #Anthropomorphism #AIHype #SiliconMirage #JulianWhatley