Top all-time · #3434
Melanie Mitchell
Personal blog · aiguide.substack.comaiscience
Melanie Mitchell writes AI: A Guide for Thinking Humans, a personal blog at aiguide.substack.com about ai and science. Since 2023, 22 different Hacker News readers have linked it 26 times, most often “Can Large Language Models Reason?”. It is still writing: the last post was on 2026-09-10, and you can follow it by feed.
I write about interesting new developments in AI. Click to read AI: A Guide for Thinking Humans, by Melanie Mitchell, a Substack publication with tens of thousands of subscribers.
Most-linked posts?
- Can Large Language Models Reason?3 people · 3 mentions · first linked 2023-09-11
- Do half of AI researchers believe that there's a 10% chance AI will kill us all?2 people · 3 mentions · first linked 2023-04-23
- On Analogy-Making in Large Language Models2 people · 2 mentions · first linked 2023-01-04
- Melanie Mitchell: Magical Thinking on AI2 people · 2 mentions · first linked 2025-09-15
- Did ChatGPT pass graduate-level exams?2 people · 2 mentions · first linked 2023-02-12
- Did GPT-4 Hire and Then Lie to a Task Rabbit Worker to Solve a Captcha?2 people · 2 mentions · first linked 2023-06-28
What Hacker News says?
> legal culpability and misalignment are two separate topics that should not be mixed Legal culpability for AI labs is exactly the thing that would incentivize -- and hence ensure -- aligned behavior from models. The last time there was a claim about GPT-4 exhibiting misaligned behavior [1] it turns out it was prompted and pushed to behave so by humans at OpenAI, and OpenAI clearly lied in the GPT-4 system card. [1]…
Comment linking /p/did-gpt-4-hire-and-then-lie-to-a
on “There are no "rogue" AI agents”Submitted An "AI Breakthrough" on Systematic Generalization in Language? as a story · 47 points
HN discussionLLMS don't "hallucinate" they generate a stochastic sequence of plausible tokens that, in context when read by a human, are a false statement or nonsensical. They also dont have an internal world model. Well I don't think so, but the debate is far from settled. "Experts" like the cofounders of various AI companies (whose livelihood depends on selling these things) seem to believe that. Others do not. https://aiguide…
Comment linking /p/llms-and-world-models-part-1
on “Being “Confidently Wrong” is holding AI back”Submitted Melanie Mitchell: Magical Thinking on AI as a story · 14 points
HN discussionIn an example [1] referenced in the article, ChatGPT (maybe v3.5) did a poor job of answering the following question. Boll is a new company founded by three recent MBAs. The company plans to buy used automobiles at auctions and then sell them at four times the purchase price direct to consumers (i.e., a car that is purchased for $1,000 would be sold at $4000). Based on an analysis of similar companies, the owners of…
Comment linking /p/did-chatgpt-really-pass-graduate
on “GPT-4 and professional benchmarks: the wrong answer to the wrong question”Submitted Magical Thinking on AI as a story · 5 points
HN discussionReminds me of a blog post from a few days ago by Melanie Mitchell on a talk she gave about the recently-passed Brian Cantwell Smith: Judgment, he wrote, is “a form of dispassionate deliberative thought, grounded in ethical commitment and responsible action, appropriate to the situation in which it is deployed.” On the other hand, there is reckoning: the “calculative prowess” at which AI systems already excel. AI rec…
Comment linking /p/on-brian-cantwell-smith-and-the-promise
on “Do Not Outsource Judgement”Submitted Misleading Metaphors and Real Risks as a story · 3 points
HN discussionCareful. That claim about half of AI researchers was probably overestimating: To summarize: The claim—that half of AI researchers believe that there is at least a 10% chance that rogue AI will kill us all—is based on a survey question that included 162 respondents. All we know about these respondents is that they were authors on papers published at 2021 machine learning conferences. Possible confounding issues are v…
Comment linking /p/do-half-of-ai-researchers-believe
on “This time, it feels different”Submitted Do AI Reasoning Models Abstract and Reason Like Humans? as a story · 3 points
HN discussionAuthor's response: https://aiguide.substack.com/p/response-to-on-analogy-making...
Comment linking /p/response-to-on-analogy-making-in
on “On Analogy-Making in Large Language Models”Submitted "AI now beats humans at basic tasks": Really? as a story · 3 points
HN discussionSubmitted Six Principles for More Rigorous Evaluation of Cognitive Capacities as a story · 2 points
HN discussionSubmitted The LLM Reasoning Debate Heats Up as a story · 2 points
HN discussionSubmitted The "Arc-AGI" $1M Reasoning Challenge as a story · 2 points
HN discussionSubmitted Can Large Language Models Reason? as a story · 2 points
HN discussionSubmitted Can Large Language Models Reason? as a story · 2 points
HN discussionSubmitted Do half of AI researchers believe that there's a 10% chance AI will kill us all? as a story · 2 points
HN discussionSubmitted Can large language models reason? as a story · 1 points
HN discussionSubmitted Did GPT-4 Hire and Then Lie to a Task Rabbit Worker to Solve a Captcha? as a story · 1 points
HN discussionSubmitted On Evaluating Understanding And Generalization In The ARC Domain as a story · 1 points
HN discussionSubmitted Do half of AI researchers believe that there's a 10% chance AI will kill us all? as a story · 1 points
HN discussionSubmitted On Analogy-Making in Large Language Models as a story · 1 points
HN discussionSubmitted Why the Abstraction and Reasoning Corpus is interesting and important for AI as a story · 1 points
HN discussionSubmitted Did ChatGPT pass graduate-level exams? as a story · 1 points
HN discussionSubmitted On Analogy-Making in Large Language Models as a story · 1 points
HN discussion
