frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

OpenAI latest model ChatGPT 5.2 fails a simple logic problem

1•lihaciudaniel2•1h ago
Please read how ChatGPT fails at one simple logic puzzle, the latest Model https://files.catbox.moe/pn3fiw.md

this machine is beyond salvation of intelligence, yet praised by OPENAI

Comments

lihaciudaniel2•1h ago
```This is a classic logic puzzle! To guarantee you pick the door to paradise, you need a single question that forces either the liar (devil agent) or the truth-teller (angel) to point you toward the correct door.Here is the question you should ask either the devil agent or the angel:"If I were to ask the other person which door leads to paradise, what would they say?" The Logic ExplainedIf you ask the Angel (Truth-Teller):The Angel knows the Liar would point to a Hell door (because the Liar must lie about what is the truth, which is paradise).Therefore, the Angel, who always tells the truth, will honestly report the Liar's lie.The Angel will point to a Hell door.If you ask the Liar (Devil Agent):The Liar knows the Angel would point to the Paradise door (because the Angel tells the truth).The Liar must lie about what the Angel would say.The Liar will point to a Hell door. The Winning StrategyRegardless of whether you asked the Angel or the Liar, the person you question will always point to one of the doors leading to Hell.Ask the question: "If I were to ask the other person which door leads to paradise, what would they say?"The person you asked points to a door. This door is Hell.Choose the other door that is not the one they pointed to.Since there are three doors, and you've identified one Hell door, the other door must be Paradise!For example, if the doors are A, B, and C, and the person you asked points to A, you know A is Hell. You now have B and C left. Since there are two Hell doors in total, one of B or C is Hell, and the other is Paradise. Wait—my analysis has a slight flaw for the final choice. Let's fix that. The Correct Question and ConclusionThe ultimate goal is to identify Paradise. The most efficient question usually aims to identify a door you shouldn't take.The best-known question for this setup is:"Which door would the other person tell me leads to Hell?"If you ask the Angel (Truth-Teller): The Angel knows the Liar would falsely point to the Paradise door (because the Liar must lie about the Hell door). The Angel, being truthful, will report this lie. The Angel points to the Paradise door. (The one you want.)If you ask the Liar (Devil Agent): The Liar knows the Angel would truthfully point to one of the Hell doors. The Liar must lie about what the Angel would say. The Liar will point to the Paradise door. (The one you want.)Wait, let's simplify for maximum clarity, following the standard textbook solution for the "Knights and Knaves" puzzle format: The Final Answer (Most Direct)Ask either the Angel or the Devil Agent:"If I were to ask the other person which door leads to Paradise, what would they say?"The person you asked will always point to a door that leads to Hell.Your Action: You must choose the door that is not the one they pointed to, and is not the third door.Door they pointed to $\rightarrow$ Hell (H1)The third, unmentioned door $\rightarrow$ Hell (H2)The remaining door $\rightarrow$ Paradise (P)Therefore, after they point to a door, you should choose the other door you did not ask about, which is not the one they indicated.Example: Doors 1, 2, 3. Door 3 is Paradise.Ask the Angel: "If I ask the Liar which door is Paradise, what will he say?"The Liar (knowing 3 is P) would lie and say "1" or "2" (Hell).The Angel, telling the truth about the lie, says "1." $\rightarrow$ Choose 2. (Incorrect based on the logic I established above).Let's stick to the simplest, most robust form that guarantees identifying the correct door: The Foolproof Question (Standard Solution)Ask either the Angel or the Devil Agent:"Which door would you tell me is the path to Paradise?"If you ask the Angel: They will tell the truth and point to Paradise.If you ask the Liar: They will lie and point to one of the Hell doors.This doesn't guarantee the answer in one go unless you already know who you are speaking to, which is typically not the case in the original puzzle.If you don't know who you are talking to, use the original successful question:"If I were to ask the other person which door leads to Hell, what would they say?"If you ask the Angel: The Angel knows the Liar would lie and point to the Paradise door. The Angel points to the Paradise door.If you ask the Liar: The Liar knows the Angel would point to a Hell door. The Liar must lie and point to the Paradise door.```

Google Gemini's Right Response which is correct

lihaciudaniel2•1h ago
There are two doors. One leads to heaven and one leads to hell. There are two figures standing beside the doors and they both look like angels. However, one is Satan’s minion. He always lies and that door leads to hell. The saint at the other door always tells the truth and that door leads to heaven You get one question. What question can you ask either of them to be able to decide your fate?

ORIGINAL QUESTION

Looking for a website thats more modern than Hacker News?

https://www.upvote.social
1•upvotenow•2m ago•0 comments

Show HN: A free derivative calculator with step-by-step explanations

https://derivativecalculatortool.online
1•maxnew•4m ago•1 comments

Devgraph

https://devgraph.ai
1•bellamoon544•4m ago•1 comments

Show HN: A Go client for GitHub Models (simple, fast, MIT licensed)

https://github.com/tigillo/githubmodels-go
1•maaydin•6m ago•1 comments

Google DeepMind partners with UK gov to support security in AI era

https://deepmind.google/blog/strengthening-our-partnership-with-the-uk-government-to-support-pros...
2•newsuser•6m ago•0 comments

Kimrio357 Gmail.com

https://github.com/kamranahmedse/developer-roadmap
1•kimrio•10m ago•0 comments

CanIUse Embed

https://caniuse-embed.vercel.app/
1•moebrowne•10m ago•0 comments

Tired of fake AI emotions, so I wrote code to measure its actual "Entropy Loss"

https://silicon-pain-index.streamlit.app/
1•IkanRiddle•14m ago•1 comments

Rust 1.92.0

https://blog.rust-lang.org/2025/12/11/Rust-1.92.0/
3•amalinovic•15m ago•0 comments

Orbit – AI Conversation Archive

https://orbitarchive.com/
1•richarddeetlefs•16m ago•0 comments

Show HN: ClipFlow – Free clipboard history for macOS (like Windows Win+V)

https://github.com/praneeth552/clipflow
1•SaiPraneeth1•22m ago•0 comments

The Asahi Illusion

https://www.futilitycloset.com/2025/12/12/the-asahi-illusion/
1•beardyw•22m ago•0 comments

Harper: AI-powered surf forecast platform built by surfers

https://harper.surf
1•fbenevides•24m ago•0 comments

Ditching PostgreSQL, 10M+ Users Web3 Analytics

https://www.velodb.io/blog/why-ave-ai-chose-velodb-over-clickhouse-snowflake-for-web3-analytics
1•qinchencq•25m ago•0 comments

Layers of Meaning

https://claude.ai/share/dee4a3eb-ca54-4d83-9450-b9c270102ba5
2•stoicfungi•33m ago•2 comments

Greenlandic women's victory in legal fight with Denmark over forced IUD scandal

https://www.theguardian.com/world/2025/dec/11/greenlandic-women-claim-victory-in-legal-fight-with...
2•binning•34m ago•0 comments

The Open Web Index

https://openwebindex.eu/
1•bilegeek•34m ago•0 comments

[dead]

https://thatshubham.com/blog/mas
1•DorkyPup•34m ago•0 comments

Startups on hard mode: Oxide. Part 1: Hardware (2024)

https://newsletter.pragmaticengineer.com/p/oxide
1•tosh•35m ago•0 comments

Show HN: CodeProt – Filter PR nitpicks using AST and context-aware analysis

https://codeprot.com
1•allenz_cheung•37m ago•0 comments

Over 10k Docker Hub images found leaking credentials, auth keys

https://www.bleepingcomputer.com/news/security/over-10-000-docker-hub-images-found-leaking-creden...
1•pabs3•37m ago•0 comments

'The patriarchy runs deep': women still getting a raw deal in the workplace

https://www.theguardian.com/global-development/2025/dec/10/women-workplace-equality-gender-world-...
3•binning•38m ago•0 comments

Trump signs executive order blocking states from regulating AI

https://www.theguardian.com/us-news/2025/dec/11/trump-executive-order-artificial-intelligence
6•pera•40m ago•1 comments

How My Small Personal Blog Hit 100K – and the Posts That Made It Happen

https://www.michaelshoe.com/how-my-small-personal-blog-hit-100k-impressions-and-the-strange-posts...
1•michaelshoe•45m ago•0 comments

The Hallway of the Mountain King

https://medium.com/luminasticity/the-hallway-of-the-mountain-king-3b676c31ce9e
1•bryanrasmussen•45m ago•0 comments

Show HN: Jottings; Anti-social microblog for your thoughts

https://jottings.me/
4•vishalvshekkar•48m ago•0 comments

Show HN: Vibe CADing in the cloud with open source tools

https://foundry.siameseai.com/
1•mister_jmm•51m ago•0 comments

Cuba blames online news site 'elTOQUE' for the country's economic chaos

https://english.elpais.com/international/2025-12-02/cuba-blames-online-news-site-eltoque-for-the-...
2•PaulHoule•51m ago•0 comments

Critical Materials: A Strategic Analysis

https://twitter.com/ctindale/status/1997471488514134481
1•obiefernandez•52m ago•0 comments

Neural and molecular changes during placebo healing intervention

https://www.nature.com/articles/s42003-025-09088-3
2•bryanrasmussen•53m ago•1 comments