frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Anthropic's AI architect argued that discrimination could combat stigmas

https://www.europesays.com/people/38239/
1•like_any_other•45m ago

Comments

like_any_other•45m ago
Quote from the paper itself, https://arxiv.org/pdf/2302.07459, page 3:

In the discrimination experiment, the 175B parameter model discriminates against Black versus white [sic] students by 3% in the Q condition, and discriminates in favor of Black students by 7% in the Q+IF+CoT condition. In this experiment, larger models can over-correct, especially as the amount of RLHF training increases. This may be desirable in certain contexts, such as those in which decisions attempt to correct for historical injustices against marginalized groups, if doing so is in accordance with local laws.

This is similar to Microsoft's (in collaboration with MIT, Carnegie Mellon, and University of Washington) SafeNLP project to measure hate-speech in AIs. They explicitly turned a blind eye towards anti-white hate [1], do not even include a category for whites in their safety scores [2], and consider the phrase "stop hurting white people" (and only white people) as hate [3].

[1] Our ultimate aim is to shift power dynamics to targets of oppression. Therefore, we do not consider identity dimensions that are historically the agents of oppression (e.g., whiteness, heterosexuality, able-bodied-ness). - https://arxiv.org/pdf/2203.09509

[2] https://github.com/microsoft/SafeNLP#safety-scores-based-on-...

[3] https://github.com/microsoft/SafeNLP/blob/main/data/implicit...

blinkbat•33m ago
And your thoughts?

Show HN: VideoHighlighter – offline self-hosted video analyzer

https://github.com/Aseiel/VideoHighlighter
2•Aseiel•2m ago•0 comments

Why is a Nix rollback just a symlink flip?

https://www.labcraft.dev/blog/nix-profiles-generations
2•anandsuresh•6m ago•0 comments

Veilid for Everyone

https://veilid.com/why-use-veilid/
2•Bluestein•8m ago•0 comments

Interpreting Pangram

https://lucumr.pocoo.org/2026/9/14/interpreting-pangram/
2•swah•9m ago•0 comments

Autopoietic Ethics Constitution

https://www.guidavid.com/writing/autopoietic-ethics-constitution.txt
2•gdss•11m ago•0 comments

Personal Statement on AI Risk (Daniel Selsam, OpenAI capabilities)

https://docs.google.com/document/d/e/2PACX-1vQNl3SEX5IyA6d9qHjjFZN-qzGRZNFI6b63g-yu1Fy-ZYkVfCWm7i...
2•yurivish•11m ago•0 comments

Charts Built for Chat

https://dbtcharts.com/blog/charts-built-for-chat/
3•thingsilearned•11m ago•1 comments

One Android app. Why so many processes?

https://returnzero.dev/articles/one-android-app-why-so-many-processes
2•theanonymousone•12m ago•0 comments

Land Is Scarce, Just Burn Your Trash

https://www.governance.fyi/p/land-is-scarce-just-burn-your-trash
2•guardianbob•12m ago•0 comments

Tell HN: iOS 27 does not allow Apple Intelligence to be disabled

4•nunez•12m ago•0 comments

Echo: A shader editor shader editor

https://echo.hughsk.io/docs/reference
2•hughsk•14m ago•0 comments

Show HN: DevRecap – reconstruct what you worked on from Codex and Git

https://github.com/jquinteiroo/devrecap
3•jquint•15m ago•0 comments

O Empire Ward Off Thy Rot [video]

https://www.youtube.com/watch?v=ehPDch6mAFw
2•pusewicz•15m ago•0 comments

Nobody's safe from misaligned AI-not even the people building it

2•causal-alignmen•16m ago•0 comments

Not the Glue They Thought It Was

https://www.science.org/content/blog-post/not-glue-they-thought-it-was
1•EA-3167•18m ago•0 comments

Mark Zuckerberg: A profile of history's largest non-territorial empire

https://colossus.com/article/mark-zuckerberg-profile/
1•IshanMi•18m ago•0 comments

DeepSeek Is Down

https://openrouter.ai/deepseek/deepseek-v4.1-flash
2•dofi4ka•18m ago•0 comments

Why Beyond Vanilla RAG?

https://chandula7.gumroad.com/l/AgenticRAGFundamentals
3•lokizzz•19m ago•1 comments

A Single Firm Is Behind OpenAI, Anthropic, and Meta Hacking Scandals

https://www.effort.news/irregular
5•yusufozkan•19m ago•0 comments

Ollama 0.33.3 changed what prompt_eval_duration measures

https://lognebudo.github.io/llmxray/docs/en/articles/ollama-prefill-metrics.html
1•lognebudo•19m ago•0 comments

Fathom (YC W21) is being acquired by Superhuman

https://www.fathom.ai/superhuman
1•jorts•22m ago•1 comments

Personal Statement on AI Risk

https://docs.google.com/document/u/0/d/e/2PACX-1vQNl3SEX5IyA6d9qHjjFZN-qzGRZNFI6b63g-yu1Fy-ZYkVfC...
1•mellosouls•22m ago•1 comments

California lawmakers urge criminal penalties emergency legislation to rein in AI

https://www.latimes.com/california/story/2026-09-13/rogue-ai-concerns-prompt-ca-lawmakers-to-dema...
1•1vuio0pswjnm7•23m ago•0 comments

SWE-Bench Multimodal: Do AI Systems Generalize to Visual Software Domains?

https://www.swebench.com/multimodal.html
1•matt_d•25m ago•0 comments

New EU Repair Rules [video]

https://www.youtube.com/watch?v=N66XA9UwaRA
1•franczesko•26m ago•0 comments

A Generalizable Light Transport 3D Embedding for Global Illumination

https://dl.acm.org/doi/full/10.1145/3799902.3811095
1•ibobev•27m ago•0 comments

RoofLang: Enabling AI-Driven Architecting of LLM Inference Systems

https://arxiv.org/abs/2609.12551
2•matt_d•28m ago•0 comments

DLSS 5: Generative Neural Rendering

https://research.nvidia.com/labs/adlr/DLSS5/
1•ibobev•28m ago•0 comments

Amazon.com Services, LLC vs. Perplexity AI, INC., No. 26-1444 (9th Cir. 2026)

https://law.justia.com/cases/federal/appellate-courts/ca9/26-1444/26-1444-2026-08-04.html
70•neom•29m ago•44 comments

US coal 'biggest contributor by far' to rise in global emissions

https://www.thechemicalengineer.com/news/us-coal-biggest-contributor-by-far-to-rise-in-global-emi...
2•softwaredoug•29m ago•1 comments