Image created with Gemini. Image prompt: A flat acrylic-style illustration of a small figure in a yellow swimsuit suspended mid-dive above a turquoise swimming pool, where the pool is assembled from a grid of slightly offset and overlapping square photo panels showing the water at different moments with hand-drawn white ripple squiggles and a not-yet-formed splash, surrounded by terracotta pink deck and lawn green grass under flat noon sunlight, hard-edged shapes, saturated flat color, no shadows, no text.
I think it is really worth reading this piece on RSI at Anthropic. There is a bit of navel-gazing, some marketing, and a lot of very sincere beliefs about what Anthropic thinks is likely in the near future of AI that you probably want to be aware of.
https://x.com/emollick/status/2062582362194460698
Policy on the AI Exponential \ Anthropic
https://www.anthropic.com/policy-on-the-ai-exponential
When AI builds itself \ Anthropic
https://www.anthropic.com/institute/recursive-self-improvement
much better ChatGPT memory:
https://x.com/gdb/status/2062608071411540196
We’ve been researching new ways for ChatGPT memory to carry context across conversations and keep it useful over time. Today, that work is rolling out as a more capable memory system in ChatGPT.
https://x.com/OpenAI/status/2062567556524003631
Mythos 5 speeds up its own training by over 69 times a human expert needs 4-8h eq. for a mere 4x speedup
https://x.com/scaling01/status/2064392809293939119
AI is advancing at a pace our policymaking institutions were never built for–and the gap between the two is becoming the central challenge of the technology. In his latest essay, our CEO Dario Amodei lays out how to close it. We’re launching three new initiatives to support the
https://x.com/AnthropicAI/status/2064783418844762489
Dario Amodei — Policy on the AI Exponential
https://darioamodei.com/post/policy-on-the-ai-exponential
One interesting pattern with Fable 5 is that it will often say things that are gibberish when I use it for coding. Things like “”The morning’s slim-scan fix cured the scan hang””, “”this is a latent-drift API-shape wrinkle””, etc. When I ask why it does this, Fable explains that it
https://x.com/tamaybes/status/2065147305494450248
The word “cancer” is flagged as a biosecurity risk by Claude Fable 5! I also tried to code a website on cancer mutations & Fable 5 was immediately removed from my list! @AnthropicAI will probably soon ban me for such dangerous prompts! FYI @karpathy “little trigger happy Fable”
https://x.com/DeryaTR_/status/2064414826122866707
New blog post: on the million-x sample efficiency gap between AIs and humans, and whether it matters: “”The reason it is relatively easy for open source and previous laggards to catch up to within months of the frontier is that data is the real driver of progress. And data can
https://x.com/dwarkesh_sp/status/2064047458708308024
NEW: malware developers added nuclear & biological weapons text to to their spyware. Goal? To trigger LLM safety refusals… so that their spyware wouldn’t be analyzed by an AI security scanner. Cleanest practical example I can think of for why over-indexing on first order
https://x.com/jsrailton/status/2064661778978533571
DeepMind cofounder Shane Legg thinks that search is essential for a model to be genuinely creative. Pre-trained base models can do incredible things. But Shane thinks this is just a matter of them mixing together existing concepts from their training data. If he’s right, coming
https://x.com/dwarkesh_sp/status/2064827030273945676
Human beings whose emotional centres are damaged, even if their intelligence is still intact, have terrible decision-making skills. Whatever role emotions are playing in humans, it’s necessary for agency. Ilya speculates that the equivalent for AIs is something to do with value
https://x.com/dwarkesh_sp/status/2062564414142972041
Imagine building a computer and not allowing its use in CS research. Thats some dystopian shit.
https://x.com/martin_casado/status/2064727048460058937
A real problem with feeling the acceleration viscerally is that current models are really good and it is hard to feel the vibe difference on most individual tasks with new models, even as AIs continue to increase in ability by large amounts (which they actually are doing).
https://x.com/emollick/status/2062573152547279042
friends don’t let friends use vector search without guardrails
https://x.com/rishdotblog/status/2065026144903315545
One reason you want AIs to be better writers is that there is a lot of writing even in software, and it is incredibly painful to hit a menu which is filled with Claudisms or ChatGPTish phrases. A report is not “”what leaves the room”” & analyses are not “”every number makes a mark””
https://x.com/emollick/status/2063368660798898284
OpenClaw legends at oversubscribed event at GutHub @steipete @davemorin @vincent_koc @OmarShahine
https://x.com/TheTuringPost/status/2062343319716512003
The best thing you could be spending time on this week is how you can avoid model lock-in. Strategize on how you can leverage different types of models. Not a choice IMO. There is no stronger data point than what we got this week on where things are headed. You want
https://x.com/omarsar0/status/2064753171214299209
Leave a Reply