En 30 secondes
Matthew Berman shows two ChatGPT responses to a question about a seahorse emoji. The responses cycle through corrections, and one continues generating before eventually presenting a fish as its answer.
- 1A simple questionBerman asks whether a seahorse emoji exists. The response says yes but repeatedly revises its proposed answer.
- 2A longer loopA second example runs for a long time and stops; Berman selects continue generating.
- 3An incorrect final answerThe displayed final answer labels a fish as a seahorse.
À retenir
- The displayed response repeatedly proposes and retracts emoji as a seahorse.
- A second response continues for a long time and requires a request to continue generating.
- The example illustrates an inconsistent answer, not a systematic assessment of ChatGPT.
Lire le texte remanié, chapitre par chapitre
The seahorse question
Berman shows a response that repeatedly corrects itself after being asked about a seahorse emoji. It never settles cleanly on an answer.
I think I just broke ChatGPT’s brain. I asked whether there is a seahorse emoji, and it said yes before listing emojis and repeatedly correcting itself: “Wait, no,” then “hold on,” then another supposedly final answer. It just keeps going.
A longer response
Another example keeps generating through a long series of corrections. Its displayed final answer calls a fish a seahorse.
That was actually one of the tamer examples. Another response went on for so long that it stopped while still working through the question. I hit “continue generating,” and it carried on in the same confused fashion. Its final answer identifies an emoji as a seahorse, but it is a fish. Did we just break ChatGPT?
Sur la même chaîne
34:21EN
Matthew Berman· 27 oct. 2026Matthew Berman Examines Anthropic’s IPO Filing, AI Growth and GovernanceMatthew Berman, Sam Altman · Forward FutureThe presenter reviews the S-1’s financial metrics, compute strategy and voting structure while questioning the tension between rapid AI development and safety.
0:36ENMatthew Berman· 25 oct. 2026Kimi K3.1 and a Rubik’s Cube Solver DemonstrationMatthew Berman · Forward FutureMatthew Berman shows a scrambled virtual cube being solved and gives the Kimi K3.1 test a favorable assessment.
25:50ENMatthew Berman· 24 oct. 2026Kimi K3.1 Pro Tested: Strong 3D Demos, Uneven Results and Competitive PricingMatthew Berman · Forward FutureMatthew Berman tests the open-weight model on coding, 3D rendering and visual explanations, then compares benchmark results and prices.
32:20ENMatthew Berman· 23 oct. 2026Who Is Local AI For? The Costs, Trade-offs and a Hybrid FutureMatthew Berman · Forward FutureMatthew Berman weighs data control and independence against hardware cost, setup complexity and inference speed.
1:03:37ENMatthew Berman· 23 oct. 2026Matthew Berman’s Team Tests Kimi K3.1 on Animation, Games, and ChartsMatthew Berman, Alex, Brian · Forward FutureA live team AMA turns into an informal test of Kimi K3.1, from personal-agent workflows to motion graphics, a browser game, and a benchmark chart.
0:53ENMatthew Berman· 22 oct. 2026Sam Altman on AI and the Discovery of New KnowledgeSam Altman · OpenAIAltman expects progress in mathematical discovery to extend to biology, materials and energy.