/Tech3h ago

Tim Rocktäschel, UCL AI professor, says recursive self-improvement and agent training require open-ended, safe environments

Researchers noted safety limitations in current AI speedrun designs.

11300672
elie
Tim Rocktäschel#68
Original post
Tim Rocktäschel@_rockt#68inTech

@eliebakouch @kellerjordan0 @karpathy Thanks @eliebakouch! Really impressed with what you do at @PrimeIntellect as well 🚀

I agree, finding the right open-ended & safe environments is going to be key to develop this technology further.

elie@eliebakouch

very cool results!

i really believe speedruns by @kellerjordan0 @karpathy and others are a great testbed for RSI like tasks, both for evaluating and improving how models do research

there are still limitations in how current speedruns are designed, and work needed on the system side to make agents efficient. seeing what models can do in math/code, i believe one missing piece to achieve the same for ai research is giving them the right environment

7:08 AM · Jun 11, 2026 · 504 Views
Sentiment

Users signal respect and approval for researchers stressing the need for open-ended safe environments in AI research.

Pos
100.0%
Neg
0.0%
1 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS171LIKES3
elie@eliebakouch

@_rockt @kellerjordan0 @karpathy @PrimeIntellect 🫡🫡

Tim Rocktäschel@_rockt

@eliebakouch @kellerjordan0 @karpathy Thanks @eliebakouch! Really impressed with what you do at @PrimeIntellect as well 🚀

I agree, finding the right open-ended & safe environments is going to be key to develop this technology further.

2hViews 171Likes 3Bookmarks 0