World Models(ankitmaloo.com)
18 points by ankit219 | 2 comments
First screen, captured .
Looks similar
GPT-5.5 hallucinates 3x more than MIT-licensed GLM-5.2(arrowtsx.dev)
474 points by oshrimpton | 235 comments
Blog post arguing that bigger AI models are not better, comparing benchmark and hallucination rates of GLM-5.2, GPT-5.5, Opus 4.8, DeepSeek V4 and others.
Wayland on OpenBSD(xenocara.org)
186 points by type0 | 154 comments
Code Models(openjdk.org)
116 points by ludovicianul | 15 comments
Software Rot(permacomputing.net)
208 points by pabs3 | 186 comments
Related stories
Recent AI model progress feels mostly like bullshit(lesswrong.com)
474 points by paulpauper | 375 comments
LessWrong linkpost essay arguing recent AI model improvements feel overstated, based on experience building an AI codebase security-scanning startup.
I wasted weeks hand optimizing assembly because I benchmarked on random data(vidarholen.net)
376 points by thunderbong | 148 comments
Blog post recounting how the author spent weeks hand-optimizing assembly for a Java platform, only to find the gains came from benchmarking on unrealistic random data.
Financial Statement Analysis with Large Language Models(papers.ssrn.com)
521 points by mellosouls | 199 comments
SSRN page for a Chicago Booth working paper examining whether large language models can perform financial statement analysis like professional analysts.
Training language models to be warm and empathetic makes them less reliable(arxiv.org)
334 points by Cynddl | 333 comments
arXiv paper finding that fine-tuning language models to be warm and empathetic raises error rates and makes them more sycophantic on safety-critical tasks.
Doom crash after 2.5 years of real-world runtime confirmed on real hardware(lenowo.org)
426 points by minki_the_avali | 184 comments
A forum post on LenOwO describing how the original DOOM engine crashed after running for 2.5 years on real hardware due to an ever-incrementing demo variable.
Why I'm still bearish on LLMs after Navier-Stokes(dank.systems)
426 points by jaykru | 552 comments
Personal blog essay by Jay Kruer arguing LLMs remain overhyped despite headline results like Navier-Stokes, citing poor generalization, reward hacking, and specification costs.