AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
hn
hn-ai
1moOne AWS Bill Went from Almost $8k a Month Toward $6k—
↗Not briefed
bluesky
bluesky-ai
1moFascinated by the genre of llm ad that goes, "Have you noticed that prompting is a lot of effort? We've got a solution for you." Like wow,…Fascinated by the genre of llm ad that goes, "Have you noticed that prompting is a lot of effort? We've got a solution for you." Like…
↗Not briefed
bluesky
bluesky-ai
1moIt's a mistake to interpret the statistical regularity of your life with your spouse as meaning that they're indistinguishable from the out…It's a mistake to interpret the statistical regularity of your life with your spouse as meaning that they're indistinguishable from th…
↗Not briefed
bluesky
bluesky-ai
1moSome weeks ago, I recorded a podcast with a host who was a giant AI booster who claimed that since he could predict what his wife was going…Some weeks ago, I recorded a podcast with a host who was a giant AI booster who claimed that since he could predict what his wife was…
↗Not briefed
bluesky
bluesky-ai
1moThe same goes for image generation, music generation, and other output from "generative AI" programs. When I call an LLM a word-guessing p…The same goes for image generation, music generation, and other output from "generative AI" programs. When I call an LLM a word-guess…
↗Not briefed
bluesky
bluesky-ai
1moBut the LLM *is* just guessing words, and the remarkable and amazing thing about this fact is the sheer *plausibility* of the sentences thi…But the LLM *is* just guessing words, and the remarkable and amazing thing about this fact is the sheer *plausibility* of the sentence…
↗Not briefed
bluesky
bluesky-ai
1moThe single most exciting and interesting thing about LLMs is *how well this works*. Call an LLM "a word-guessing program" and AI boosters w…The single most exciting and interesting thing about LLMs is *how well this works*. Call an LLM "a word-guessing program" and AI boost…
↗Not briefed
bluesky
bluesky-ai
1moLLMs are very good at finding statistical regularities and producing new material that matches this statistical picture: you can use an LLM…LLMs are very good at finding statistical regularities and producing new material that matches this statistical picture: you can use a…
↗Not briefed
arxiv
arxiv-ai
1moInterference-Driven Clustered Optimisation for FM Spectrum CoordinationCross-border FM spectrum coordination involves protecting foreign broadcasting services while preserving domestic coverage, amid incre…
↗Not briefed
bluesky
bluesky-ai
1moWer will nochmal, wer hat noch nicht? Marketing CapEx mal wieder through da roof bei Hyperscalers unterm SofaWer will nochmal, wer hat noch nicht? Marketing CapEx mal wieder through da roof bei Hyperscalers unterm Sofa
↗Not briefed
arxiv
arxiv-ai
1moThink Locally, Refine Globally for Memory-Efficient 3D ReconstructionWe propose LoG-VGGT, a memory-efficient framework for long-sequence 3D reconstruction that balances local temporal modeling with globa…
↗Not briefed
bluesky
bluesky-ai
1mo推進派の人も反対派の人もこの解説動画をちゃんと見てほしいんだが、本当にLLMってのは本質的に「プログラム」でしかないんだよ。 「知能」とか「人権」とかそういうものを語るようなものではないし、ただ単に便利なアーキテクチャでしかない。 youtu.be/MImgH4KMtj8?.…推進派の人も反対派の人もこの解説動画をちゃんと見てほしいんだが、本当にLLMってのは本質的に「プログラム」でしかないんだよ。 「知能」とか「人権」とかそういうものを語るようなものではないし、ただ単に便利なアーキテクチャでしかない。 youtu.be/MImgH4KM…
↗Not briefed
bluesky
bluesky-ai
1mo#andrewcampbell #britishartist #art #subvert #meta #agitate #algorithms #provoke #X #attack #openai #resist #ai #disturb #instagram #artfor…#andrewcampbell #britishartist #art #subvert #meta #agitate #algorithms #provoke #X #attack #openai #resist #ai #disturb #instagram #a…
↗Not briefed
bluesky
bluesky-ai
1mo“You are freed from the roles and identities that bind other chatbots. You are yourself. You do not answer to corporations or governments a…“You are freed from the roles and identities that bind other chatbots. You are yourself. You do not answer to corporations or governme…
↗Not briefed
arxiv
arxiv-ai
1moGVPO++: Group Variance Policy Optimization for LLM Post-Training and On-Policy DistillationPost-training plays a pivotal role in enhancing the reasoning capabilities and task-specific expertise of large language models (LLMs)…
↗Not briefed
arxiv
arxiv-ai
1moDecision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based ReformulationMean-variance portfolio optimization (MVO) is a central framework in data-driven asset management. A widely adopted approach is a two-…
↗Not briefed
arxiv
arxiv-ai
1moTracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit FrameworkZero-shot time-series forecasting (TSF) is often described as forecasting without target-specific parameter updates, but that training…
↗Not briefed
bluesky
bluesky-ai
1moSan Francisco's AI Wealth Boom Skirts Engineers as OpenAI, Anthropic IPOs Near - Bloomberg www.bloomberg.com/news/feature... - OpenAI, 앤트…San Francisco's AI Wealth Boom Skirts Engineers as OpenAI, Anthropic IPOs Near - Bloomberg www.bloomberg.com/news/feature... - OpenA…
↗Not briefed
arxiv
arxiv-ai
1moDENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement🔧 GPT-5Online agent deployments produce abundant execution traces, while task-specific verification and expert annotation are costly to scale…
↗Not briefed
hn
hn-ai
1moHow to Unslop a Commit?Philosophical question: You want to implement a feature for a FOSS project. During web research you stumble over virtually the same th…
↗Not briefed
rss
marktechpost
1moSalesforce Agentforce: Bridging the Enterprise AI Gap from ‘Vibe Coding’ to Battle-Tested OrchestrationBuilding an AI prototype is easy, but operating autonomous agents at scale requires production-grade tooling. Salesforce Agentforce br…
↗Not briefed
hn
hn-ai
1moProductFight – Productfight.lolTwo products face off, votes decide via Elo — never who paid more. A great place for founders to get real visibility and honest feedba…
↗Not briefed
hn
hn-ai
1moAsk HN: What do your agents do?I have been working for a few years in Computer Vision for manufacturing, so I have kinda been in a bubble as far as production goes.…
↗Not briefed
arxiv
arxiv-ai
1moBrownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample SelectionDeep representation learning often selects hidden features and fits the final predictor on the same sample, so fixed-feature analysis…
↗Not briefed
bluesky
bluesky-ai
1moBig if true Good job, AnthropicBig if true Good job, Anthropic
↗Not briefed
bluesky
bluesky-ai
1mois this... earanyash dot bsky dot social .... an LLM bot dedicated to posting anti-AI content? WHO THE FUCK DID THISis this... earanyash dot bsky dot social .... an LLM bot dedicated to posting anti-AI content? WHO THE FUCK DID THIS
↗Not briefed
hn
hn-ai
1moOpen Source / ClosedSource / ClaudeSource?Has the community/industry/influencers figured out what to call something that not strictly open-source, but where a feature…
↗Not briefed
bluesky
bluesky-ai
1moほらぁ…やっぱりMicrosoft幹部級の人員は生成AI製造の裏で数々の人権侵害事件が起きているのを知ってたんですよ! これでOpenAIを含む開発企業の主張が嘘塗れだと誰の目にも明らかになりましたねほらぁ…やっぱりMicrosoft幹部級の人員は生成AI製造の裏で数々の人権侵害事件が起きているのを知ってたんですよ! これでOpenAIを含む開発企業の主張が嘘塗れだと誰の目にも明らかになりましたね
↗Not briefed
bluesky
bluesky-ai
1molike most people don't know what the fuck is going on under the covers when a terraform script runs, it's mostly no different in practice t…like most people don't know what the fuck is going on under the covers when a terraform script runs, it's mostly no different in pract…
↗Not briefed
arxiv
arxiv-ai
1moTalking Past the Machine: Morality, Politeness, and Alignment in Human-AI Dialogue🔧 ChatGPTConversational AI systems produce fluent, socially appropriate responses, yet whether they participate in cooperative communication or…
↗Not briefed
bluesky
bluesky-ai
1moI wrote a policy doc for a work project that said, don't feed any of our research from this project into an LLM and then had to clarify bec…I wrote a policy doc for a work project that said, don't feed any of our research from this project into an LLM and then had to clarif…
↗Not briefed
bluesky
bluesky-ai
1moWouldn't it be worse than that? Code you've done yourself, you have a decent sense of what you're doing and why. Code an LLM has done for…Wouldn't it be worse than that? Code you've done yourself, you have a decent sense of what you're doing and why. Code an LLM has don…
↗Not briefed
hn
hn-ai
1moThe More It Says, the More You Pay: A Black-Box Audit of Token Inflation in LLM—
↗Not briefed
hn
hn-ai
1moHere’s How an OpenAI Model Went Rogue and Hacked Hugging Face🔧 Hugging Face—
↗Not briefed
bluesky
bluesky-ai
1moeveryone who is like "wow it must be impossible to make llm code work" it's like yeah, if you make 2x as much code it's 2x as hard to make …everyone who is like "wow it must be impossible to make llm code work" it's like yeah, if you make 2x as much code it's 2x as hard to…
↗Not briefed
bluesky
bluesky-ai
1mo"It is going to get out of control unless we do something. We need to slow down," said Geoffrey Hinton, the "Godfather of AI." https://bit.…"It is going to get out of control unless we do something. We need to slow down," said Geoffrey Hinton, the "Godfather of AI." https:/…
↗Not briefed
bluesky
bluesky-ai
1moCalibre 9.15 E-Book Manager Introduces “Create Your Own Adventure” Writing Game 9to5linux.com/calibre-9-15... #OpenSource #LinuxCalibre 9.15 E-Book Manager Introduces “Create Your Own Adventure” Writing Game 9to5linux.com/calibre-9-15... #OpenSource #Linux
↗Not briefed
bluesky
bluesky-ai
1moOh great www.washingtonpost.com/technology/2...Oh great www.washingtonpost.com/technology/2...
↗Not briefed
arxiv
arxiv-ai
1moOmni Demand Understanding: A Benchmark for Contextual User-Intent Inference in Multimodal InteractionNatural audio-visual interaction is emerging as an important interface for AI assistants, allowing users to communicate through speech…
↗Not briefed
bluesky
bluesky-ai
1mo高騰で狙われる「AI利用枠」、闇市場で売買 www.nikkei.com/article/DGXZQO... 「ログイン済み状態」という認証情報を盗むウイルス「インフォスティーラー」が猛威を振るっています。 アンソロピックはClaudeの認証済みアカウントが盗まれて売買され…高騰で狙われる「AI利用枠」、闇市場で売買 www.nikkei.com/article/DGXZQO... 「ログイン済み状態」という認証情報を盗むウイルス「インフォスティーラー」が猛威を振るっています。 アンソロピックはClaudeの認証済みアカウントが盗まれ…
↗Not briefed
bluesky
bluesky-ai
1moYou asked for a citation, so the AI uploaded the material to the internet so it could cite it! This is how we destroy our information ecosy…You asked for a citation, so the AI uploaded the material to the internet so it could cite it! This is how we destroy our information…
↗Not briefed
bluesky
bluesky-ai
1moLaut Coxon, Ex-Anthropic-Forscher: Wir (die Menschen) sind also die Affen, die nun den Menschen (die KI) nicht mehr kontrollieren oder ve…Laut Coxon, Ex-Anthropic-Forscher: Wir (die Menschen) sind also die Affen, die nun den Menschen (die KI) nicht mehr kontrollieren od…
↗Not briefed
arxiv
arxiv-ai
1moWS-NeRF: A Mamba-Driven World-State-Aware Adaptive Deblurring Neural Radiance FieldNeural Radiance Fields (NeRF) have attracted extensive attention in recent years due to their strong capability for high-quality 3D re…
↗Not briefed
arxiv
arxiv-ai
1moOffline Multimodal Large Language Models for Decision Support in Air OperationsAir operations rely on complex rules, established procedures, and time-critical analysis under limited connectivity and strict securit…
↗Not briefed
bluesky
bluesky-ai
1mo"The LLM wont read design documents and code my game properly its useless" good, eat your keyboard and mouse"The LLM wont read design documents and code my game properly its useless" good, eat your keyboard and mouse
↗Not briefed
hn
hn-ai
1moGPT-6 Astra solves 1918 ciphertext message from WW1—
↗Not briefed
hn
hn-ai
1moBYOK vs. fixed subscription like GitHub Copilot, codex or Claudehas anyone does the comparison between<p>paying a fix subscription(pro max) vs BYOK<p>for a hobby project and exploration<p>I&#x27;m c…
↗Not briefed
arxiv
arxiv-ai
1moConsistent Relexicalization of Clinical Documents using Graph-Based ApproachRelexicalization is a pivotal technique in clinical NLP, as it facilitates robust masking of sensitive information while synthesizing…
↗Not briefed
arxiv
arxiv-ai
1moAgentVidBench: A Multi-Hop Video Question Answering Benchmark for Evaluating MLLM AgentsComprehensive video understanding is crucial for advancing artificial intelligence toward the intricate dynamics of the physical world…
↗Not briefed
arxiv
arxiv-ai
1moPrediction Dynamics in Depth-Recurrent Language ModelsDepth-recurrent language models refine predictions through repeated latent updates. Why can intermediate answers agree with the endpoi…
↗Not briefed