Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Xiaomi MiMo v2.6 Pro: New best open-weight LLM with simple design(sebastianraschka.com)
    discuss
  2. Can Phones Explain Pisa?(commonreader.co.uk)
    discuss
  3. Adaptive reranking with Jev and a logistic regression(prospex.ch)
    discuss
  4. Undetectable Conversations Between AI Agents via Noise-Resilient Key Exchange(arxiv.org)
    discuss
  5. Show HN: A free help center launch kit – 20 customizable pages(knowledgebase.dev)
    discuss
  6. Data liberation: Apache Kafka's native cluster mirroring(redhat.com)
    discuss
  7. Seashells, Software, by the Seashore(charliemeyer.co)
    discuss
  8. KDE at 30: What Are the Colored Blobs Telling Us?(ipsquad.net)
    discuss
  9. Apple's first live music venue is opening beneath its UK headquarters(theverge.com)
    discuss
  10. Botzilla – Web automation using visual scripting(botzilla.dev)
    1comments
  11. Show HN: Methadone – Sieze the means of production from our agentic overlords(github.com/xophmeister)
    discuss
  12. Bioresorbable batteries for transient ingestible bioelectronics(nature.com)
    discuss
  13. Anton Repponen(repponen.com)
    discuss
  14. Physics in the Age of LLMs(ozamram.substack.com)
    discuss
  15. US Military Modifies AI and Lethal Targeting Process After Iran School Strike(bloomberg.com)
    1comments
  16. Show HN: ParseRail, production AI endpoints behind one key and a credit wallet(thecompound.tech)
    discuss
  17. Does a Three-Dimensional Android Dream of Four-Dimensional Sheep?(medium.com/artem-x)
    discuss
  18. ArrowSpace Performance Analysis: 100K Dimensions on Dorothea(tuned.org.uk)
    discuss
  19. Type fm in the terminal on macOS 27
    discuss
  20. Show HN: Nowly – An open-source alternative to PreMiD for Discord Rich Presence(nowly.me)
    discuss
  21. IAS Scholar Reveals Universal Rule of Black Hole "Burps"(ias.edu)
    discuss
  22. Blink – A 452KB decision model in C/WASM with no token generation(github.com/sqliteai)
    discuss
  23. Helix: The tool powering Shopify's native migration(shopify.engineering)
    discuss
  24. Luna Delegation for Codex(gist.github.com)
    discuss
  25. Laya, an open decision model that plays Tetris by itself(brainfunctioncollapse.com)
    discuss
  26. Alexa+ now generates podcast episodes on demand(aboutamazon.com)
    discuss
  27. A Chrome extension that blurs AI slop on X and LinkedIn(github.com/davertor)
    discuss
  28. Every Single-Page App Is a Worse Browser Running Inside a Browser(hypermedia.lol)
    discuss
  29. Jevlang: Policy Engine for LLM Decisions(jevlang.sh)
    discuss
  30. 2DWillNeverDie(2dwillneverdie.com)
    discuss

People Training OpenAI's AI Fired for Using AI to Train the AI

13 pointsby 56m ago404media.co
4 comments
16m agoHN ↗

So OpenAI is not a believer in recursive self-improvement after all?

13m agoHN ↗

Bubble bursting behavior in action - AI company that pushes everyone to use AI, replace everything and everyone with AI, fires people for using AI to make the AI.

The more you think about it, the worse it gets.

11m agoHN ↗

Makes sense, they are buying distillation of human brains not their own weights.

10m agoHN ↗

Typically, for AI training, you need human feedback. If AI were enough, then OpenAI would do it themselves. Though we all know how bad AI slop is, running it in a loop can lead to unreadable sentences.

They hired contractors with the condition to provide human feedback without AI; those people broke the rules, so their contracts ended prematurely.

This is more a failure of journalistic understanding than poor lab behavior.