We Don't Have to Be This Bad at Improving Society
Article URL: https://kasperjunge.com/blog/we-dont-have-to-be-this-bad-at-improving-society/ Comments URL: https://news.ycombinator.com/item?id=48758242 Points: 12 # Comments: 2
Full feed
Feed 27,897 posts
Every story across every category, newest first. Each card links to the original publisher; daily-brief posts open as editorial pages.
RT shreya rajpal: Re @swyx with the 🐐 takes on making a talk rejigged my whole talk around the thesis slide for AIE world fair
RT shreya rajpalRe @swyx with the 🐐 takes on making a talkrejigged my whole talk around the thesis slide for AIE world fairswyx @aiDotEngineer WF: lots…
The xiaoren are not giving up! DeepSeek sees itself as a company that is building AGI. What has changed was the scale and the maturity of the AI stack…
The xiaoren are not giving up!DeepSeek sees itself as a company that is building AGI. What has changed was the scale and the maturity of the…
Personalization as Inverse Planning: Learning Latent Design Intents for Agentic Slide Generation via Structural Denoising
Slide design requires personalizing both deck themes and page layouts. Yet, current AI agent-based methods struggle with fine-grained, page-level design. Solely relying on prespecified templates or…
PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception
We introduce PerceptionRubrics, a rubric-based evaluation framework that addresses the gap between saturated benchmark scores and real-world brittleness. Shifting evaluation from holistic semantic matching to rigorous…
NoPA: Non-Parametric Online 3D Scene Graph Generation
Classic 3D scene graph generation approaches fail to work in real-time due to the heavy computational cost of environment mapping and the need to generate intermediate…
The Fall of the Theorem Economy
Article URL: https://davidbessis.substack.com/p/the-fall-of-the-theorem-economy Comments URL: https://news.ycombinator.com/item?id=48758048 Points: 6 # Comments: 0
RT _horror: Its far worse, boring, banal yet also evil. Captcha solver powered boomer immortality, global KYC compliance capitalism, voluntary demogra…
RT _horrorIts far worse, boring, banal yet also evil. Captcha solver powered boomer immortality, global KYC compliance capitalism, voluntary demographic extinctionismReclaim The Net: Democrats have chosen…
Another Update to Agentic AI on Windows XP (SimpleLLMChat 1.2.5)
Application Screenshot I made some updates to my agentic harness targeting legacy machines (Windows XP on .NET 4.0), figured I'd share them. Reasoning level is now…
The jobs market is insane these days
The jobs market is insane these daysShawn Presser: I’ll be actually-homeless soon (living out of my car) if I can’t land a job. Our house is…
Very strange For what it's worth I agree that Opus 4.6 and Fable feel qualitatively similar whereas Opus 4.8 is different (and often disgusting). No i…
Very strangeFor what it's worth I agree that Opus 4.6 and Fable feel qualitatively similar whereas Opus 4.8 is different (and often disgusting). No idea how…
There are two hypotheses for the DeepSeek-V4's strange performance (as in, V4-Flash is about as good as we expected, but V4-Pro is disappointing given…
There are two hypotheses for the DeepSeek-V4's strange performance (as in, V4-Flash is about as good as we expected, but V4-Pro is disappointing given its scale):1)…
I’m selling my M1 128gb.
Hi yall, just wanted to share that - i'm out. I had a blast reading through most of the threads daily, i waited with you hoping…
Palantir CEO rages against closed models
For context, this week they struck a deal to buy Nvidia chips and run local models for their enterprise clients. So in this video he is…
Multimodal Continuous Reasoning via Asymmetric Mutual Variational Learning
Multimodal Large Language Models (MLLMs) are often constrained by a language-space bottleneck, forcing complex visual reasoning into discrete tokens which can lose perceptual nuance. A promising…
ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving
In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models…
A cheap trick for reliable structured output: feed the validation error back into the retry
If you generate structured output from an LLM and validate against a schema, you know the failure mode: usually fine, occasionally a missing field or an…
MarketFish – Simulate a market with 128 AI consumers before you launch
Article URL: https://github.com/Key-wxh/market-fish Comments URL: https://news.ycombinator.com/item?id=48757500 Points: 6 # Comments: 0
unlikely alliances form people discover previously unthinkable ideological priorities The world unites against the Avengers level threat
unlikely alliances formpeople discover previously unthinkable ideological prioritiesThe world unites against the Avengers level threatRicardo: Palantir's CEO just exposed Sam Altman and Dario Amodei for robbing…
The gauge broke: devs felt 20% faster with AI, measured 19% slower
Article URL: https://intrepidkarthi.com/writing/the-gauge-broke/ Comments URL: https://news.ycombinator.com/item?id=48757440 Points: 15 # Comments: 1
How do you all decide which quantization format to use for deployment? (built a CLI to compare them, curious if others do this differently)
Every time I need to deploy a model I end up manually testing 3-4 quant formats to see what actually holds up on my hardware VRAM,…
AI Safety Is Testing the Wrong Environment
The Lab Problem in AI SafetyThere's something that doesn't sit right with me about where alignment research happens.So much research, so many researchers, ideas, experiments, but…
qwen 3.6 35a3b on 32gb ram + 8gb vram?
is it possible to run a reasonable quant of qwen 3.6 (or I guess even ornith/3.5) 35B A3B on a laptop with 32gb of lpddr5 (7500…
Extremely interested in where Etched goes. There are good engineers on both sides of the debate, and it's unusually polarized. the "MFU-thermals" angl…
Extremely interested in where Etched goes. There are good engineers on both sides of the debate, and it's unusually polarized. the "MFU-thermals" angle does sound like…