News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
bluesky bluesky-ai | 11d | It's like the protein folding problems in biochem. Just because we know how what proteins are and how they fold doesn't mean we know all th… | It's like the protein folding problems in biochem. Just because we know how what proteins are and how they fold doesn't mean we know a… | ↗Not briefed |
bluesky bluesky-ai | 11d | You are correct. The devs are trying to build a machine. The LLM has access to all the documentation and every public codebase and "knows" … | You are correct. The devs are trying to build a machine. The LLM has access to all the documentation and every public codebase and "kn… | ↗Not briefed |
hn hn-ai | 11d | Saltline: First-person shooter built by Fable 5.2 | — | ↗Not briefed |
arxiv arxiv-ai | 11d | Abstention Is Not Threshold Compliance: Stated Scoring Rules in Open Language Models | Kalai et al. propose stating the scoring rule in the prompt with an explicit threshold t, so that a model should answer only when its… | ↗Not briefed |
hn hn-ai | 11d | LLMs Are Different Now | — | ↗Not briefed |
bluesky bluesky-ai | 11d | I'm holding off on doing these discussions because so many of these AI companies are so close to going bankrupt. I see a near future where … | I'm holding off on doing these discussions because so many of these AI companies are so close to going bankrupt. I see a near future w… | ↗Not briefed |
bluesky bluesky-ai | 11d | main takeaway from the whole IGF anti-AI discussion should be that genAI has permeated almost all code and art creation under capitalism th… | main takeaway from the whole IGF anti-AI discussion should be that genAI has permeated almost all code and art creation under capitali… | ↗Not briefed |
hn hn-ai | 11d | Silencing Watchdogs: India's Crackdown on Those Documenting Online Censorship | — | ↗Not briefed |
bluesky bluesky-ai | 11d | In conversations I often encounter LLM advocates saying "You may not like it, but the one thing you can't deny is [thing that is incredibly… | In conversations I often encounter LLM advocates saying "You may not like it, but the one thing you can't deny is [thing that is incre… | ↗Not briefed |
arxiv arxiv-ai | 11d | Closing the Loop: Co-Evolving EM for Irregular Time Series Generation in Lifted Representations | Regular time-series generation from irregular time series is typically reduced to one-shot imputation: complete each partial sequence… | ↗Not briefed |
bluesky bluesky-ai | 11d | I feel like the people who are up-in-arms about this don’t understand how programming actually works Before we copy+pasted from Claude, we… | I feel like the people who are up-in-arms about this don’t understand how programming actually works Before we copy+pasted from Claud… | ↗Not briefed |
bluesky bluesky-ai | 11d | people who are against *any* LLM usage in software are really funny to see like what, are those couple lines of code are going to “poison … | people who are against *any* LLM usage in software are really funny to see like what, are those couple lines of code are going to “po… | ↗Not briefed |
hn hn-ai | 11d | Show HN: I have trained 15 blood glucose forecasting models on my own data | — | ↗Not briefed |
arxiv arxiv-ai | 11d | Thompson Sampling using Prior-fitted Diffusion Transformers | Thompson sampling is natural randomized strategy for optimizing unknown functions in a black-box manner. In continuous global optimiza… | ↗Not briefed |
bluesky bluesky-ai | 11d | straight up: i do not like the way people treat AI/LLM use as being akin to sin and damnation. i am firmly against AI and avoid it whenever… | straight up: i do not like the way people treat AI/LLM use as being akin to sin and damnation. i am firmly against AI and avoid it whe… | ↗Not briefed |
hn hn-ai | 11d | Woolly, a local memory shared by all your AI tools | — | ↗Not briefed |
arxiv arxiv-ai | 11d | Cleaning an Agent's History Hurts When the World Still Holds the Error | Tool-using language-model agents accumulate histories of their own mistakes, and a system that compacts, repairs or hands off such a h… | ↗Not briefed |
bluesky bluesky-ai | 11d | Big Walk is my game of the year, without contention It being excluded from selection as game of the year based off LLM usage is a wholly v… | Big Walk is my game of the year, without contention It being excluded from selection as game of the year based off LLM usage is a who… | ↗Not briefed |
bluesky bluesky-ai | 11d | PSI is trying to make a business out of AI-driven scientific discovery | PSI is trying to make a business out of AI-driven scientific discovery | ↗Not briefed |
hn hn-ai | 11d | Battery prediction pilot: GPT Astra 3/12 correct choices; baseline 5/12 | — | ↗Not briefed |
rss theverge-ai | 11d | The US is winning the AI race. Jensen Huang could have blown it | As espionage targets go, Jensen Huang is perhaps one of the least surprising on Earth. The Nvidia CEO runs one of the world's most val… | ↗Not briefed |
rss techcrunch-ai | 11d | PSI is trying to make a business out of AI-driven scientific discovery | PSI used an unreleased Anthropic model to answer a fundamental question about quantum computing that scientists have pursued for nearl… | ↗Not briefed |
arxiv arxiv-ai | 11d | Multiple-Choice Secretary Problem with Predictions | We study the multiple-choice secretary problem, in which candidates arrive in uniformly random order and the goal is to select at most… | ↗Not briefed |
bluesky bluesky-ai | 11d | timesofindia.indiatimes.com/world/us/a-6... | timesofindia.indiatimes.com/world/us/a-6... | ↗Not briefed |
bluesky bluesky-ai | 11d | Like the problem with any discussion of LLM usage across the board is that it quickly devolves into becoming ideologically impure by contac… | Like the problem with any discussion of LLM usage across the board is that it quickly devolves into becoming ideologically impure by c… | ↗Not briefed |
bluesky bluesky-ai | 11d | ‘Anthropic revealed key details about its potentially record-breaking IPO on Tuesday, giving gave more fodder to critics of the rapidly gro… | ‘Anthropic revealed key details about its potentially record-breaking IPO on Tuesday, giving gave more fodder to critics of the rapidl… | ↗Not briefed |
arxiv arxiv-ai | 11d | Evaluating the Performance of LLMs in Drafting NIH Data Management Plans | Sharing research data is essential to promote transparency, enable reproducibility, and fuel scientific discoveries. To ensure consist… | ↗Not briefed |
bluesky bluesky-ai | 11d | Building well-behaved AI one model at a time isn’t enough. As AI agents begin collaborating, safety will depend on the very world they shar… | Building well-behaved AI one model at a time isn’t enough. As AI agents begin collaborating, safety will depend on the very world they… | ↗Not briefed |
bluesky bluesky-ai | 11d | I cannot pretend to care about the big walk llm shit | I cannot pretend to care about the big walk llm shit | ↗Not briefed |
arxiv arxiv-ai | 11d | Evaluating AI Musical Creativity: Towards a Quantitative Multi-Dimensional Framework Based on Value, Novelty, and Surprise | Despite the increasing sophistication of generative Artificial Intelligence (AI), research on the evaluation of AI-generated musical c… | ↗Not briefed |
arxiv arxiv-ai | 11d | Medical Image Segmentation: Foundation Model Adaptation versus Task-Specific Training | Foundation models pretrained on large natural-image collections are increasingly adapted for medical image segmentation, in the expect… | ↗Not briefed |
bluesky bluesky-ai | 11d | So apparently, the EU’s latest “historic” sanctions package against Russia is riddled with errors: defunct companies, incorrect names... I… | So apparently, the EU’s latest “historic” sanctions package against Russia is riddled with errors: defunct companies, incorrect names.… | ↗Not briefed |
arxiv arxiv-ai | 11d | Toward Foundation Detection Models via Example-Based Grounding in EM-DETR | DeepLearning (DL)-based Computer-Aided Diagnostic (CAD) systems have shown great promise in supporting clinicians, but their performan… | ↗Not briefed |
youtube youtube-aidailybrief | 11d | Nvidia CEO Admits Cheating with Suspected Chinese Spy | Today we have some real spy novel stuff as the CEO of Nvidia admits to having an affair with a woman the US government believes is a C… | ↗Not briefed |
arxiv arxiv-ai | 11d | TorchCurves: Differentiable Parametric Curves | We present TorchCurves, a PyTorch library that treats vector-valued parametric curves as learnable primitives. Its central operation m… | ↗Not briefed |
arxiv arxiv-ai | 11d | AutoScaffold: AI-Guided Scaffolding for Reinforcement Learning of Language Models | Reinforcement learning with verifiable rewards (RLVR) produces rich training experience, yet outcome rewards reduce each attempt to a… | ↗Not briefed |
arxiv arxiv-ai | 11d | Contrastive Discovery: Open-Ended Scientific Discovery over Competing Explanations | LLM agents are increasingly used for open-ended, data-driven scientific discovery, yet most of them evaluate hypotheses pointwise, tes… | ↗Not briefed |
bluesky bluesky-ai | 11d | I asked whether the model was contained and being run on a local server such that my data wouldn’t be shared outside their local system. Sh… | I asked whether the model was contained and being run on a local server such that my data wouldn’t be shared outside their local syste… | ↗Not briefed |
bluesky bluesky-ai | 11d | At annual checkup today my PCP informed me her clinic was now using an AI model to do visit summaries. I asked if she was being forced to u… | At annual checkup today my PCP informed me her clinic was now using an AI model to do visit summaries. I asked if she was being forced… | ↗Not briefed |
bluesky bluesky-ai | 11d | Not going to lie, there's a part of me hoping the LLM replies back "I'm sorry, Ursula, I'm afraid I can't do that." just to keep up with … | Not going to lie, there's a part of me hoping the LLM replies back "I'm sorry, Ursula, I'm afraid I can't do that." just to keep up… | ↗Not briefed |
bluesky bluesky-ai | 11d | The only good LLM is a dictionary. The only good data center is a library. | The only good LLM is a dictionary. The only good data center is a library. | ↗Not briefed |
hn hn-ai | 11d | I built a Claude Code mod that turns my English prompts into a learning log | — | ↗Not briefed |
bluesky bluesky-ai | 11d | Companies like Anthropic or OpenAI are, in Wark's terminology, *vectoralists.* They are moving to create/control chokepoints in the flows o… | Companies like Anthropic or OpenAI are, in Wark's terminology, *vectoralists.* They are moving to create/control chokepoints in the fl… | ↗Not briefed |
hn hn-ai | 11d | Show HN: Totpvault - a universal desktop app for your TOTPs | — | ↗Not briefed |
bluesky bluesky-ai | 11d | Some prior reporting with my esteemed colleagues on all this: OpenAI reporter Ashley Capoot and I covered the Musk v. Altman trial extensiv… | Some prior reporting with my esteemed colleagues on all this: OpenAI reporter Ashley Capoot and I covered the Musk v. Altman trial ext… | ↗Not briefed |
bluesky bluesky-ai | 11d | Someone on Twitter used a paid AI to "compose a sonata in the style of early Beethoven" and I'm really, really, really interested by all of… | Someone on Twitter used a paid AI to "compose a sonata in the style of early Beethoven" and I'm really, really, really interested by a… | ↗Not briefed |
rss latent-space | 11d | NaiveAI Promises 2,100 Tokens per Second. How Well Does It Code? | NaiveAI could potentially top Artificial Analysis’s speed rankings. Our hands-on test suggests its coding quality holds up against Dee… | ↗Not briefed |
arxiv arxiv-ai | 11d | Pyrite: Building and Benchmarking Protein-Ligand Pose Ranking Agents | Predicting how a ligand binds to its target is central to small-molecule drug discovery. Modern cofolding models can generate diverse… | ↗Not briefed |
hn hn-ai | 11d | Anthropic Has No Moat | — | ↗Not briefed |
bluesky bluesky-ai | 11d | like once again not a coder but isn’t finding ways to solve a problem or circumvent an issue kind of like… the primary joy of doing it?? is… | like once again not a coder but isn’t finding ways to solve a problem or circumvent an issue kind of like… the primary joy of doing it… | ↗Not briefed |