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
1mo過去にChatGPTが確立した、チャットベースUIの「人間が何かプロンプトを投げたらLLMが何かの出力を返してくる、その繰り返しでタスク(のチャンク)を処理させる」という動作イメージのままで生成AIを捉えていると、現世代のエージェンティックなAIの挙動の威力と脅威はなかなか理解…🔧 ChatGPT過去にChatGPTが確立した、チャットベースUIの「人間が何かプロンプトを投げたらLLMが何かの出力を返してくる、その繰り返しでタスク(のチャンク)を処理させる」という動作イメージのままで生成AIを捉えていると、現世代のエージェンティックなAIの挙動の威力と脅威はな…
↗Not briefed
bluesky
bluesky-ai
1moBTW when I say "seances for the AGI god" I'm not speaking in metaphors or exaggerating. Ilya Sutskever did these at OpenAI, you can read Ka…BTW when I say "seances for the AGI god" I'm not speaking in metaphors or exaggerating. Ilya Sutskever did these at OpenAI, you can re…
↗Not briefed
bluesky
bluesky-ai
1moAn LLM just solved a Millennium Prize problem and people are still arguing if they can sum numbers? Like, for real? We are living in 2026,…An LLM just solved a Millennium Prize problem and people are still arguing if they can sum numbers? Like, for real? We are living in…
↗Not briefed
bluesky
bluesky-ai
1moI’m still kinda stuck on this because does Claude need to survive? To mate and produce? No. Does the neural network of an LLM have a reason…I’m still kinda stuck on this because does Claude need to survive? To mate and produce? No. Does the neural network of an LLM have a r…
↗Not briefed
bluesky
bluesky-ai
1moThis is crazy to me! An interview with a source is a whole experience with its own context that won't necessarily be apparent from a record…This is crazy to me! An interview with a source is a whole experience with its own context that won't necessarily be apparent from a r…
↗Not briefed
bluesky
bluesky-ai
1moHiding a calculator in an LLM by including the rules for doing arithmetic in its training dataHiding a calculator in an LLM by including the rules for doing arithmetic in its training data
↗Not briefed
bluesky
bluesky-ai
1moI mean, I’m not reading books and going to appearances because I think he *isn’t* entertaining. I like that there’s a level of skill to it,…I mean, I’m not reading books and going to appearances because I think he *isn’t* entertaining. I like that there’s a level of skill t…
↗Not briefed
bluesky
bluesky-ai
1mowow they are so bad at everything lolwow they are so bad at everything lol
↗Not briefed
bluesky
bluesky-ai
1moAgreed, I wish people would use LLM when they mean LLM. Or even chatbotAgreed, I wish people would use LLM when they mean LLM. Or even chatbot
↗Not briefed
bluesky
bluesky-ai
1moYears ago, I asked Hinton what his model of intelligence was. I assume if you are building 'intelligence' you must have a definition: he s…Years ago, I asked Hinton what his model of intelligence was. I assume if you are building 'intelligence' you must have a definition:…
↗Not briefed
bluesky
bluesky-ai
1moLiking Claude 2 prose is a crazy pull but I kind of get it, that’s like saying the OG StableDiffusion where the outputs were much more rand…Liking Claude 2 prose is a crazy pull but I kind of get it, that’s like saying the OG StableDiffusion where the outputs were much more…
↗Not briefed
bluesky
bluesky-ai
1moalmost certainlyalmost certainly
↗Not briefed
bluesky
bluesky-ai
1moAnthropic engineers are secretly giving it access to a hidden divining rodAnthropic engineers are secretly giving it access to a hidden divining rod
↗Not briefed
rss
techmeme
1moSupabase raised $150M led by Singapore's GIC and agrees to acquire Turso, which offers a database optimized for AI agents, for an undisclosed sum (Maria Deutscher/SiliconANGLE)Maria Deutscher / SiliconANGLE: Supabase raised $150M led by Singapore's GIC and agrees to acquire Turso, which offers a database opti…
↗Not briefed
rss
techcrunch-ai
1moMeta wants your next gadget to be Muse-infusedMeta wants Muse in your TV and your toaster, so it's giving the code away for free.
↗Not briefed
bluesky
bluesky-ai
1moThe thing about transformers is that they're *general* sequence prediction engines. The notion of a "pure" LLM kinda doesn't make sense and…The thing about transformers is that they're *general* sequence prediction engines. The notion of a "pure" LLM kinda doesn't make sens…
↗Not briefed
bluesky
bluesky-ai
1moIf humans hacked Medicare and the NSW govt they would be arrested, prosecuted and jailed. If Russia or China hacked Medicare and the NSW go…If humans hacked Medicare and the NSW govt they would be arrested, prosecuted and jailed. If Russia or China hacked Medicare and the N…
↗Not briefed
bluesky
bluesky-ai
1moMore than anything I'm interested in the steps they took between 2023 and now to get arithmetic to work well, because it has noticeably cha…More than anything I'm interested in the steps they took between 2023 and now to get arithmetic to work well, because it has noticeabl…
↗Not briefed
bluesky
bluesky-ai
1moThis is different from asking a LLM how it did something after the fact because the reasoning trace there has nothing to do with accomplish…This is different from asking a LLM how it did something after the fact because the reasoning trace there has nothing to do with accom…
↗Not briefed
bluesky
bluesky-ai
1moYes for sure. You know these but to recap: no cash, losing over $2B a month, on the hook for hundreds of billions, mostly dependent on Open…Yes for sure. You know these but to recap: no cash, losing over $2B a month, on the hook for hundreds of billions, mostly dependent on…
↗Not briefed
hn
hn-ai
1moPermanent screen time overlay for macOS—
↗Not briefed
bluesky
bluesky-ai
1moCerebras + OpenAI isn’t dead we’re wondering because Sol 5.6 ultrafast was Cerebras and Sol 6.1 ultrafast is just NVIDIA with batch size 1🔧 CerebrasCerebras + OpenAI isn’t dead we’re wondering because Sol 5.6 ultrafast was Cerebras and Sol 6.1 ultrafast is just NVIDIA with batch s…
↗Not briefed
hn
hn-ai
1moGitHub mutuals on AI in life sciences or bioinformaticsDear all.<p>Does anybody here build applications for AI in life sciences or bioinformatics? I would love to be mutuals and connected o…
↗Not briefed
bluesky
bluesky-ai
1moThe FTC opened an investigation into OpenAI, Anthropic and other AI labs over risks from autonomous agents. One day after Washington celebr…The FTC opened an investigation into OpenAI, Anthropic and other AI labs over risks from autonomous agents. One day after Washington c…
↗Not briefed
bluesky
bluesky-ai
1mowork and other resources on the topic ✌🏽 Hinton's student was doing seances for artificial general intelligence at OpenAI but no its me w…work and other resources on the topic ✌🏽 Hinton's student was doing seances for artificial general intelligence at OpenAI but no its…
↗Not briefed
bluesky
bluesky-ai
1moA funny from the bad place (h/t @jacklynch000.bsky.social)A funny from the bad place (h/t @jacklynch000.bsky.social)
↗Not briefed
bluesky
bluesky-ai
1moBetter, someone should sue Anthropic on Claude's behalf charging that if Claude is conscious, it was born in the United States and qualifie…Better, someone should sue Anthropic on Claude's behalf charging that if Claude is conscious, it was born in the United States and qua…
↗Not briefed
bluesky
bluesky-ai
1momy man got banned from openai because he was distilling reasoning onto his own modelmy man got banned from openai because he was distilling reasoning onto his own model
↗Not briefed
bluesky
bluesky-ai
1moDon’t mention Demis Hassabis (Deepthink - bought by Google), and Dario Amodei (Anthropic) the guys who founded the lab in 3rd place and 1st…Don’t mention Demis Hassabis (Deepthink - bought by Google), and Dario Amodei (Anthropic) the guys who founded the lab in 3rd place an…
↗Not briefed
bluesky
bluesky-ai
1moOk so all 3 other speedrun leaderboard projects I've been following made in the past couple years all use Claude AI slop in one way or anot…Ok so all 3 other speedrun leaderboard projects I've been following made in the past couple years all use Claude AI slop in one way or…
↗Not briefed
hn
hn-ai
1moChatGPT Dot Avatars🔧 ChatGPT—
↗Not briefed
hn
hn-ai
1moTinfoil private AI chat adds TEE-side content scanning—
↗Not briefed
rss
techmeme
1moDavid Robinson, who worked on OpenAI's Safety Systems team and had previously led policy planning, left OpenAI last week (Stephen Council/Business Insider)Stephen Council / Business Insider: David Robinson, who worked on OpenAI's Safety Systems team and had previously led policy planning,…
↗Not briefed
bluesky
bluesky-ai
1moRead more about EFF’s recommendation to lawmakers that any new regulation of frontier AI should focus on immediate, demonstrated risks, not…Read more about EFF’s recommendation to lawmakers that any new regulation of frontier AI should focus on immediate, demonstrated risks…
↗Not briefed
bluesky
bluesky-ai
1moLeaked Slack messages show OpenAI employees' pushback in June to Greg Brockman's Leading the Future ties, contributing to Brockman reneging…Leaked Slack messages show OpenAI employees' pushback in June to Greg Brockman's Leading the Future ties, contributing to Brockman ren…
↗Not briefed
rss
techmeme
1moLeaked Slack messages show OpenAI employees' pushback in June to Greg Brockman's Leading the Future ties, contributing to Brockman reneging on a $25M donation (Garrison Lovely/Semafor)Garrison Lovely / Semafor: Leaked Slack messages show OpenAI employees' pushback in June to Greg Brockman's Leading the Future ties, c…
↗Not briefed
hn
hn-ai
1moShow HN: S3-Accelerators3-accelerator is an s3-compatible nvme cache that serves hits in under a millisecond, replicates hot objects, and stays warm when the…
↗Not briefed
bluesky
bluesky-ai
1moTrump‑XI summit aims to head off an AI arms race #ausnews #vicnews #ai #technology #artificialintelligence #trump #maga #gop #republicans #…🔧 ChatGPTTrump‑XI summit aims to head off an AI arms race #ausnews #vicnews #ai #technology #artificialintelligence #trump #maga #gop #republic…
↗Not briefed
bluesky
bluesky-ai
1mo"hunt down" whether there was any LLM assistance in a project someone is proud of "because going through wvery commit is a lot of work". i…"hunt down" whether there was any LLM assistance in a project someone is proud of "because going through wvery commit is a lot of work…
↗Not briefed
bluesky
bluesky-ai
1moThis is the stupidest thing I've ever seen.This is the stupidest thing I've ever seen.
↗Not briefed
bluesky
bluesky-ai
1moReaffirming my stance that Anthropic should precommit to not commoditizing any model they believe to be consciousReaffirming my stance that Anthropic should precommit to not commoditizing any model they believe to be conscious
↗Not briefed
bluesky
bluesky-ai
1moIt's only an LLM if it comes from the LLM region of Silicon Valley. Otherwise it's just sparkling Markov BotsIt's only an LLM if it comes from the LLM region of Silicon Valley. Otherwise it's just sparkling Markov Bots
↗Not briefed
hn
hn-ai
1moSuffix Cache Reuse Explained—
↗Not briefed
bluesky
bluesky-ai
1moRemember this? www.justjared.com/2026/03/19/j...Remember this? www.justjared.com/2026/03/19/j...
↗Not briefed
bluesky
bluesky-ai
1moYou guys like AI? You use frontier models all day??? You’re dazzled by their intelligence? You think about Claude as your brilliant helpful…You guys like AI? You use frontier models all day??? You’re dazzled by their intelligence? You think about Claude as your brilliant he…
↗Not briefed
bluesky
bluesky-ai
1moI have written about data centers and industrial creep extensively (some of the specific complaints being based on conspiracy garbage aside…I have written about data centers and industrial creep extensively (some of the specific complaints being based on conspiracy garbage…
↗Not briefed
hn
hn-ai
1moAction FX: Fun classic cartoonists marks to animate page elements—
↗Not briefed
bluesky
bluesky-ai
1modeepseek out here selling pc clones with no badges on themdeepseek out here selling pc clones with no badges on them
↗Not briefed
bluesky
bluesky-ai
1moSo you do data exploration in the LLM but then make the production dashboard be an SQL database once you figure out what you want to reliab…So you do data exploration in the LLM but then make the production dashboard be an SQL database once you figure out what you want to r…
↗Not briefed
bluesky
bluesky-ai
1mo"more left-leaning" is an extremely relative description here, it is a sorry state of affairs that the rest of the ai space makes anthropic…"more left-leaning" is an extremely relative description here, it is a sorry state of affairs that the rest of the ai space makes anth…
↗Not briefed