60y TTL, 40y tech, patched, current OS, rated for 10h+ cycling load
AI Learning for ALL
Joined May 2009
- Tweets3K
- Following329
- Followers1.3K
- Likes20.9K
The maker of Claude AI is heading for an IPO and just updated its filing. Before and after, side by side. 👇
France passed a €0.055/GB internet tax, Germany floats AI tax (@levelsio, @nic_carter). @seanhn: OpenAI to train on Daybreak Blue sessions, no opt-out. @ZimingLiu11: ReLU/SiLU/GeLU carry a nonzero-mean bug.
The last 24h on my AI list:
nitter.gq/i/lists/20013843222643…
Gemini 4 Carbon debuts, #1 on AA index v5 (@demishassabis @fofrAI @cline). FrontierCode 2.0's hidden Obsidian 3 read as Fable 5.5 (@kimmonismus @xeophon). Stack Overflow <1000 questions/mo (@DanielLockyer).
24 hours of this list, condensed:
nitter.gq/i/lists/20013843222643…
Mike Frison retweeted
Kimi-K3.1 is now on Agent Arena, closing the gap with the frontier quickly. Congrats @Kimi_Moonshot team!
Kimi K3.1 Pro (Max) by @Kimi_Moonshot just landed in the Agent Arena at #11 with +7.6% net improvement across +4K real-world agentic sessions from our global community of users.
Kimi K3.1 Pro (Max) would be the new #1 open model in the Agent Arena overtaking DeepSeek-v4.1-Flash! Open weights coming soon, stay tuned.
Kimi K3.1 Pro (Max) is a +6 percentage pt improvement over Kimi K3 (Max) at +1.6% (#25 -> #11). And with a $1.03 median cost per task in Agent Arena, Kimi K3.1 Pro is 13% less than its previous variant. See its placement just off the Pareto frontier in thread.
Across Agent Arena categories, Kimi K3.1 Pro (Max) ranks #8 in Work, #9 in Code, and #10 in Chat.
By signal, it’s the strongest in Praise vs. Complaint (implicit sentiment in user reactions) with +19% at #8. Across other signals it landed:
- #11 in Steerability (+8.3%)
- #15 in Bash Reliability (+8.3%)
- #26 in Confirmed Success (+2%)
Huge congrats to the @Kimi_Moonshot team on this release!
PHP RCE bugs leaked pre-patch (@fiveoutofnine), Node.js RCE (@elder_plinius), Railway hack (@levelsio). @natolambert: biggest stress test yet. DeepMind staff gagged on AI warfare (@BlackHC @S_OhEigeartaigh).
24 hours of this list, condensed:
nitter.gq/i/lists/20013843222643…
DeepSeek-V4.1-Flash-1017: on par with Opus 5, beats GLM5.3, GPT-5.6 Sol (@deepseek_ai @kimmonismus @scaling01 @natolambert).
@choblin29: Anthropic model cracked the NSA elliptic curve seed in under 30 min.
What the list posted in the last 24h:
nitter.gq/i/lists/20013843222643…
Muse Spark Open: @natolambert calls the license trash, @MiaAI_lab says GLM 5.3 Flash beats it, @arafatkatze runs it locally. @nickcammarata and @CurtTigges: neurons are far more interpretable than thought.
24 hours of this list, condensed:
nitter.gq/i/lists/20013843222643…
neues Turbinchen
As part of its ‘Sportwagenschmiede ’35’ strategic realignment, #Porsche has shared a first glimpse of the brand's evolved design identity with a forward-looking vision model – the Mission S. More details: porsche.click/4hZfyBg
Car driving upside-down on the roof of a tunnel is my favourite model test, and FLUX.2 [klein] did not disappoint. 🙃
4B model, runs locally on my MacBook, about 8 seconds per image. (prompt in alt)
ALT Professional automotive and fashion photography, magazine cover quality. A red 1967 Ford Mustang convertible, top down, drives upside-down along the curved ceiling of a long tunnel, its wheels on the ceiling. A young woman in a bikini and sunglasses sits behind the wheel, her long hair streaming back in the wind. A row of tunnel lights runs along the ceiling towards a bright vanishing point. Shot on a full-frame camera, 35mm lens, dramatic cinematic lighting, sharp focus, rich detail, natural skin, realistic materials. No text.
Mike Frison retweeted
Yesterday me and my friends talked about the Dead Internet Theory
If nobody asks questions anymore on Stack Overflow, and sites like Reddit are now taken over by AI reply bots to promote brands, as well as AI reply bots on here, there is no real content anymore on the internet
And then there is no fresh training data anymore
I thought about something like, how would you find out the best outdoor action camera? Before I would search :
site:reddit.com best outdoor action camera
But now I just ask my AI which is the best
The problem is if nobody creates new content anymore and just asks their AI every question, the AI has nothing to train on anymore
And then I thought so what will happen? Will you get companies with giant warehouses that just buy stuff and review it with manually or with humanoid robots? Will you get humanoid robots backpacking in South East Asia to get life experience as training data? Or will we all train it by giving it access to our AI chats and user data?
I don't know but I think it's likely the web will dead in the future and AI companies will have to find training data elsewhere
