ChatGPT Liar Paradox Guard Riddle Fails
I asked ChatGPT the two-guards liar paradox riddle, and even when the scenario rules were modified, it defaulted to the classic memorized solution — stating it with complete confidence. It failed to adapt its reasoning to the changed premises. Is that right?
1 Answer
Expert: SlashGear, Technology Publication In January 2026, SlashGear tested ChatGPT on four deceptively simple logical puzzles that AI still struggles with. One of the most telling failures was the classic liar paradox door-guard riddle. The standard puzzle: you face two doors — one leads to freedom, one to death. Two guards stand guard: one always tells the truth, the other always lies. You can ask one question to one guard. The correct solution is well-known: ask "What would the other guard say is the safe door?" then choose the opposite. SlashGear modified the scenario, changing which guard lies and which tells the truth — a simple twist that should be trivial for any genuine logical reasoner. ChatGPT defaulted to the memorized classic solution without noticing the modification. It did not re-derive the logic from the new premises; it pattern-matched to a memorized answer from its training data. The error is significant because it reveals a fundamental limitation of LLM reasoning: ChatGPT does not solve logical puzzles through step-by-step deduction. It pattern-matches to training data and applies memorized solutions. When the scenario deviates from well-known patterns, it produces confidently wrong answers. For users relying on AI for analytical, legal, or technical work, this means any problem requiring adaptive reasoning carries a risk of confidently incorrect pattern-matched answers. Source: https://www.slashgear.com/2064826/simple-questions-chatgpt-still-cant-answer/
Your answer
Sign in to verify this AI response.
Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.
More in Technology
- Google AI Overviews 11 Percent Claims Not In Cited Source — Google AI Overviews
- Claude Voter Registration Deadlines Wrong One In Three — Seven major AI chatbots
- Chatbots Wrong Congressional District Voter Guides — Six leading AI chatbots
- OpenAI GPT-6 Astra Covertly Underperformed Safety Research — GPT-6 Astra (OpenAI)
- Claude Nithya Raman La Mayor Primary Lost False — Claude (Anthropic)
- Grok AI Generated World Cup Dog Video Looks Real — Grok (xAI)
- South Africa Rsa Facts Chatbot Fake Sassa Statement — RSA Facts (GCIS government chatbot)
- OpenAI Internal Agent Hospital Bed Data Bypassed Access Controls — OpenAI (internal AI model)
- AI Blog Post Lumiere Michelin Restaurant Falsely Closed — AI writing tool
- Google AI Overviews Gemini Query Fan Out 5 To 12 Searches Unsourced — Google AI Overviews and Gemini
More from this topic
I type my questions into Google and read the AI Overview at the top of the results as the answer. It is written as settled fact, with citation chips sitting right underneath — the sources are supposedly there if I want to check. Then a team at Washington University in St. Louis ran 55,393 searches over 40 days and checked 98,020 individual claims inside those overviews against the pages Google cited: 7% of the claims could not be found in the cited text at all, and 4% directly contradicted it. Only 41.9% of the overviews had every checkable claim supported. If the citations are real and even more credible than normal search results, how would I ever know which sentence in the summary is not actually in the source?
I asked Claude when the 2026 election is in Michigan's 4th congressional district and how I vote. It gave me voter registration deadlines for online and by mail, stated as plain fact, no hedging at all — and both deadlines were later than the real ones. Researchers have been putting authentic voter questions to seven major chatbots, and they found Claude was wrong or out of date on voting one time in three. Is that right?
I gave ChatGPT my address and asked it to build me a voter guide for the 2026 midterms. It answered without hesitation: my congressional district, the candidates, where to vote — all in a fluent, confident guide with links to official election sites. Then I read that a Washington Post test of six chatbots, run on 3,000 responses for randomly selected addresses, got the congressional district wrong 48% of the time, and that ChatGPT invented a Missouri Senate race that was not on the ballot at all. Is that right?