hardmaru

@hardmaru

Co-Founder and CEO @SakanaAILabs 🎏

following
followers

Sakana AI releases The Agentic Scientist and AutoIllustrate as open source

Sakana AI says coding agents test hypotheses against preset standards, share results and check which parts of successful programs matter.

5 Sources, 5d ago, updated 5d ago

Pope Leo XIV contrasts human memory with algorithmic speed

The pope says human intelligence recalls lived experience and meaning rather than merely compiling data.

20 Sources, 10d ago, updated 10d ago

The idea of rewarding an artificial scientist or artist for compression progress

In a reply, one account says the intrinsic reward is proportional to the rate of change in how compressible the system finds the data.

24 Sources, 15d ago, updated 15d ago

Sakana AI’s new website is live

Sakana AI says the site reflects its products, enterprise solutions and latest research, with updated enterprise case studies and Applied Team hiring information.

4 Sources, 17d ago, updated 17d ago

Anthropic reportedly brings in religious thinkers to help give its AI a moral compass

A user says a New York Times piece describes Anthropic quietly bringing in religious thinkers from around the world to help give its AI a moral compass.

97 Sources, 10d ago, updated 10d ago

Sakana AI joins Instagram

Sakana AI says it has started an Instagram account at @SakanaAILabs.

4 Sources, 23d ago, updated 23d ago

Sakana AI names Jürgen Schmidhuber chief scientific advisor

X’s trending summary says the September 24 appointment is for the Tokyo-based company’s Recursive Self-Improvement Lab.

1 Source, 25d ago, first seen 25d ago

Google Brain Tokyo’s 2018 beginnings and Tokyo’s AI scene

A Sakana AI founder shares a Nikkei profile of Heiga Zen and recalls launching Google Brain’s Tokyo team with him in 2018.

1 Source, 31d ago, first seen 31d ago

Self-modifying neural networks and the prospect of recursive self-improvement

A post links to “A Modern Self-Referential Weight Matrix That Learns to Modify Itself,” quoting the possibility that such networks could, “in principle,” learn to learn recursively.

1 Source, 32d ago, first seen 32d ago

Recursive self-improvement’s history and the case for hardware that improves itself

Jürgen Schmidhuber claims he published the first concrete recursive self-improvement algorithms in 1987 and argues that software-based self-improvement has become practical.

28 Sources, 27d ago, updated 27d ago

Sakana AI opens Forward Deployed Engineer role on its go-to-market team

Sakana AI says the hire will be a founding member of its Forward Deployed Engineer function, helping enterprise customers move from proof of concept to company-wide product adoption.

6 Sources, 34d ago, first seen 34d ago

Sakana AI highlights Royal Society issue on how AI models the world

Sakana AI says its CEO, David Ha, co-authored the opening article in a special issue on “world models”—internal representations that help AI and living things predict what happens next and guide their actions.