


human (aka syvb/smitop)


The spend figures in this post are likely somewhat inflated by researchers using ultrafast internally, which is 6x more expensive (since there's no employee spend limits) - it would be helpful for OpenAI to say how much the fast/ultrafast price premium affected the spend figures
We're sharing new data on how AI is changing research at OpenAI. We find researchers' AI use continues to rise exponentially, and agents are completing longer tasks, but human involvement remains central in research. Researchers are also spending more time on safety and alignment, with human effort in these areas growing faster than in any other individual research area. This report aims to increase transparency around research acceleration and inform the governance of increasingly capable AI. We believe that AI labs should be required to track and publicly report indicators of recursive self-improvement (RSI): the process by which AI accelerates the development of future AI systems. Our hope is that this expanded report will help sustain and advance shared measurement norms across the industry. openai.com/index/research-ac…


(kinda inspired by x.com/voooooogel/status/2122…)
deep dreamed images that activate neurons and chain lens readouts in qwen 3.6 27b







If you find one cockroach in your freezer room, your best estimate of the number of cockroaches is apparently a bimodal distribution between "1" and "a lot"
If you find two ants in your kitchen, the best estimate of the total number of ants in your kitchen is not two




Here's a bug I reported in the UMA protocol last year (which was really minor and AFAIK problematic) that's slightly interesting: gist.github.com/syvb/65db8ef…
