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
bluesky
bluesky-ai
11dUnfortunately relenting and using LLMs simply digs a deeper hole, this is by design of the companies that are trying to turn everyone into …Unfortunately relenting and using LLMs simply digs a deeper hole, this is by design of the companies that are trying to turn everyone…
↗Not briefed
bluesky
bluesky-ai
11dI support transparency in AI the way Steam does it. But, like the Big Walk devs said, their use was so limited as to not even mandate discl…I support transparency in AI the way Steam does it. But, like the Big Walk devs said, their use was so limited as to not even mandate…
↗Not briefed
bluesky
bluesky-ai
11dbut again, we live in "autocorrect is evil because it uses a machine learning model and Big Tech calls that AI now", computer literacy is d…but again, we live in "autocorrect is evil because it uses a machine learning model and Big Tech calls that AI now", computer literacy…
↗Not briefed
bluesky
bluesky-ai
11dturning a big dial taht says "LLM usage" on it and constantly looking back at the audience for approval like a contestant on the price is r…turning a big dial taht says "LLM usage" on it and constantly looking back at the audience for approval like a contestant on the price…
↗Not briefed
arxiv
arxiv-ai
11dSeqFold: Timestamp-Addressable Compressed States for User Action Sequence BackbonesExisting sequential user modeling methods mainly focus on improving the backbone encoder or learning task-specific user representation…
↗Not briefed
bluesky
bluesky-ai
11dNuanced, but not dismissive, piece on the energy usage of various models highlighting research about *where* queries pollute most (the US!)…Nuanced, but not dismissive, piece on the energy usage of various models highlighting research about *where* queries pollute most (the…
↗Not briefed
bluesky
bluesky-ai
11dSemantic collapse and other maintenance-aggravating LLM coding issues do seem like they are a more pernicious kind of problem though.Semantic collapse and other maintenance-aggravating LLM coding issues do seem like they are a more pernicious kind of problem though.
↗Not briefed
hn
hn-ai
11dShow HN: Clingon – Markdown cheatsheets that run in your real shellClingon is a very opinionated tmux based TUI that contains a cheatsheet menu in one panel, and a live shell in another panel, all in o…
↗Not briefed
bluesky
bluesky-ai
11dgenuinely the most maddening thing to me about the llm/ai arguments is that usage of these systems equates to taking the medicine sold by t…genuinely the most maddening thing to me about the llm/ai arguments is that usage of these systems equates to taking the medicine sold…
↗Not briefed
bluesky
bluesky-ai
11di think the other take i see that makes me tilt my head is "but it was at the end, this amount isn't personally enough for me to care" i'm…i think the other take i see that makes me tilt my head is "but it was at the end, this amount isn't personally enough for me to care"…
↗Not briefed
bluesky
bluesky-ai
11dwithout making any judgement whether llm use for it is good or bad, code is just like absolutely not the same thing as visual art or writin…without making any judgement whether llm use for it is good or bad, code is just like absolutely not the same thing as visual art or w…
↗Not briefed
bluesky
bluesky-ai
11dBurke right on top of tech with the phrase 'didnt know computer was loaded' /s. Computers aren't 'loaded', software is 'loaded' in fact the…Burke right on top of tech with the phrase 'didnt know computer was loaded' /s. Computers aren't 'loaded', software is 'loaded' in fac…
↗Not briefed
bluesky
bluesky-ai
11dnothing and nobody exists in a vacuum, those tools exist in a moment and society that sucks ass at the moment. everything is nuanced imo an…nothing and nobody exists in a vacuum, those tools exist in a moment and society that sucks ass at the moment. everything is nuanced i…
↗Not briefed
bluesky
bluesky-ai
11dwas the LLM trained on stolen labour? then yes, they're "AI folks"was the LLM trained on stolen labour? then yes, they're "AI folks"
↗Not briefed
bluesky
bluesky-ai
11dthe idea that "just using claude for code assistance/checks" or "just using a chatbot for research/troubleshooting help" doesn't count as t…the idea that "just using claude for code assistance/checks" or "just using a chatbot for research/troubleshooting help" doesn't count…
↗Not briefed
arxiv
arxiv-ai
11dTopological Dynamics Under Non-Stationary Supervision: Persistent Homology of Neural Network Weight EvolutionWe investigate how the persistent topology of neuron- weight point clouds evolves under stable and non-stationary supervision. An iden…
↗Not briefed
bluesky
bluesky-ai
11dLlama 4 Maverick (17B-128E MoE) vs DeepSeek V3 (0324): Which LLM API actually protects your SaaS gross margin at 10,000 DAU? Here is the l…🔧 Llama 4Llama 4 Maverick (17B-128E MoE) vs DeepSeek V3 (0324): Which LLM API actually protects your SaaS gross margin at 10,000 DAU? Here is…
↗Not briefed
bluesky
bluesky-ai
11dThey fully explain what happened here...basically *they* looked at Unity's code and couldn't figure it out. Then *UNITY* looked at Unity's …They fully explain what happened here...basically *they* looked at Unity's code and couldn't figure it out. Then *UNITY* looked at Uni…
↗Not briefed
bluesky
bluesky-ai
11dIn regards to House House, Big Walk, and AI/LLM use, here is their FAQ page from the first of November web.archive.org/web/20261101... And …In regards to House House, Big Walk, and AI/LLM use, here is their FAQ page from the first of November web.archive.org/web/20261101...…
↗Not briefed
arxiv
arxiv-ai
11dOn Private Knowledge Retrieval and Private Inference for secure multi-agent systemsMulti-agent systems provide a convenient framework for the massive deployment of agents based on Large Language Models (LLMs). It prov…
↗Not briefed
bluesky
bluesky-ai
11dDo you really think an LLM being trained on millions of years' worth of writing, to replace millions of human jobs, is the same thing as a …Do you really think an LLM being trained on millions of years' worth of writing, to replace millions of human jobs, is the same thing…
↗Not briefed
bluesky
bluesky-ai
11dI decided to implement an AI/LLM policy for PossiblyAxolotl Co LTD Industries going forwardI decided to implement an AI/LLM policy for PossiblyAxolotl Co LTD Industries going forward
↗Not briefed
arxiv
arxiv-ai
11dDecision-Focused Learning via Neural Optimization Layers for Battery-Assisted Grid-Import Smoothing in AI Data CentersRapid AI data-center power fluctuations challenge grid integration. Sustained grid-import smoothing requires allocating limited batter…
↗Not briefed
bluesky
bluesky-ai
11dNeon silks, moonlit gaze #DigitAlart #AIArt #AIArtwork #AIGenerated #StableDiffusion #OpenAINeon silks, moonlit gaze #DigitAlart #AIArt #AIArtwork #AIGenerated #StableDiffusion #OpenAI
↗Not briefed
bluesky
bluesky-ai
11dIt’s also dumb as fuck because the search engine was highly profitable and nobody has found a way to make ads work in an LLM which is why t…It’s also dumb as fuck because the search engine was highly profitable and nobody has found a way to make ads work in an LLM which is…
↗Not briefed
bluesky
bluesky-ai
11dIf I look at the LLM summary on Google, I always check the source anyway, which kinda defeats the whole purpose. Then it's just regular goo…If I look at the LLM summary on Google, I always check the source anyway, which kinda defeats the whole purpose. Then it's just regula…
↗Not briefed
bluesky
bluesky-ai
11d1. big walk's critical flaw re: LLM usage is that it was developed without a policy in place (there was usage but it was very minimal) 2. I…1. big walk's critical flaw re: LLM usage is that it was developed without a policy in place (there was usage but it was very minimal)…
↗Not briefed
bluesky
bluesky-ai
11dPeople are going to want to find a pass for House House using AI because of a 'valid use case,' and I get it, I do. But that still ignores …People are going to want to find a pass for House House using AI because of a 'valid use case,' and I get it, I do. But that still ign…
↗Not briefed
arxiv
arxiv-ai
11dWalrus Technical ReportWe present Walrus, a unified 8B speech-and-audio foundation model handling speech recognition (including code-switching and multi-talk…
↗Not briefed
bluesky
bluesky-ai
11dThey used an LLM to help solve a performance issue in Unity, a thing that LLMs are incredibly good at. This is a very silly thing to be mad…They used an LLM to help solve a performance issue in Unity, a thing that LLMs are incredibly good at. This is a very silly thing to b…
↗Not briefed
hn
hn-ai
11dFlashStore: A Ceph OSD Rewritten in Rust for NVMe—
↗Not briefed
arxiv
arxiv-ai
11dYou Are What You Read: Indian Political News Media Bias Dataset for Hindi LanguageMedia bias understanding and detection remains an under-researched problem for Indian context, specifically for Indic languages, such…
↗Not briefed
arxiv
arxiv-ai
11dAURA: Auditory Spatial Intelligence via a Unified Representation from a General-Audio Encoder through Complete AlignmentAnswering spatial questions about multi-source scenes requires a model both to identify each sound and to locate it. Existing spatial…
↗Not briefed
arxiv
arxiv-ai
11dEgo Guidance: Enhancing Robot Policy Learning with Action-Free Human Egocentric VideosIn this paper, we investigate how human egocentric videos can benefit robot policy learning without action supervision. Building on a…
↗Not briefed
bluesky
bluesky-ai
11di do, however, want to point out the hypocrisy of part of house house's disclosure. imagine saying "we're worried about the normalization o…i do, however, want to point out the hypocrisy of part of house house's disclosure. imagine saying "we're worried about the normalizat…
↗Not briefed
bluesky
bluesky-ai
11dAI criminal penalties www.bbc.com/news/article...AI criminal penalties www.bbc.com/news/article...
↗Not briefed
bluesky
bluesky-ai
11dif claude was a little helper bot that ran locally I could see this new attitude about it being the "ethical ai", but it's not even close. …if claude was a little helper bot that ran locally I could see this new attitude about it being the "ethical ai", but it's not even cl…
↗Not briefed
rss
bing-news-ai
11dSMX Develops AI-Powered Scientific Intelligence Platform, Targets Commercial Launch in Q2 2027SMX (NASDAQ:SMX) announced the development of SMX Scientific Intelligence (SMX SCI), an artificial intelligence-powered platform desig…
↗Not briefed
arxiv
arxiv-ai
11dBootstrapped On-Policy DistillationOn-policy distillation (OPD) improves language models by distilling teacher predictions on student-generated trajectories, while on-po…
↗Not briefed
bluesky
bluesky-ai
11dto be clear i support the rules that ban LLM usage entirely. i just also don't really care that they used it to patch a bug as a last ditch…to be clear i support the rules that ban LLM usage entirely. i just also don't really care that they used it to patch a bug as a last…
↗Not briefed
rss
simon-willison
11dIntroducing Our Smallest, Most Efficient Model: Muse Ember—
↗Not briefed
rss
techmeme
11dSoftBank reports Q2 net loss of $7B vs. a ~$15.8B profit a year ago, on ~$14B revenue vs. ~$12.8B est. as tech selloff weighs on results; shares close up 13.9% (Jenny Lee/CNBC)Jenny Lee / CNBC: SoftBank reports Q2 net loss of $7B vs. a ~$15.8B profit a year ago, on ~$14B revenue vs. ~$12.8B est. as tech sello…
↗Not briefed
bluesky
bluesky-ai
11dyou know, i wouldn’t be too up in arms if the LLM usage was indeed minimal and the Big Walk studio was regretful. but the last paragraphs …you know, i wouldn’t be too up in arms if the LLM usage was indeed minimal and the Big Walk studio was regretful. but the last paragr…
↗Not briefed
arxiv
arxiv-ai
11dOpticalRAG: Pixel-Space Compression for Token-Efficient Retrieval-Augmented Generation🔧 ModalRetrieval-augmented generation (RAG) is bottlenecked by the token budget of large language models: feeding long documents into the con…
↗Not briefed
bluesky
bluesky-ai
11dLPT: using an LLM chat bot for coding issues has the same environmental impact as using it image generation. It uses the same stolen data f…LPT: using an LLM chat bot for coding issues has the same environmental impact as using it image generation. It uses the same stolen d…
↗Not briefed
bluesky
bluesky-ai
11dBig Walk released in August and they're only now disclosing LLM usage per the IGF rulesBig Walk released in August and they're only now disclosing LLM usage per the IGF rules
↗Not briefed
bluesky
bluesky-ai
11dmy takeaway here is that anyone using an LLM to write ten lines of code for their game in the future is just gonna keep their mouth shut ab…my takeaway here is that anyone using an LLM to write ten lines of code for their game in the future is just gonna keep their mouth sh…
↗Not briefed
bluesky
bluesky-ai
11d"U.S. data centers alone could consume more natural gas than Germany and Japan combined by 2035" - feck the climate 🤦‍♂️ "One analyst ex…"U.S. data centers alone could consume more natural gas than Germany and Japan combined by 2035" - feck the climate 🤦‍♂️ "One analy…
↗Not briefed
bluesky
bluesky-ai
11dUltimately a casual search from an LLM with no source access and only Unity's fairly bad documentation is not going to unveil something bet…Ultimately a casual search from an LLM with no source access and only Unity's fairly bad documentation is not going to unveil somethin…
↗Not briefed
bluesky
bluesky-ai
11dI sincerely don't know what to do about education. It appears the existence of LLMs has made education effectively impossible. I'd prefer t…I sincerely don't know what to do about education. It appears the existence of LLMs has made education effectively impossible. I'd pre…
↗Not briefed