AI researcher Nathan Lambert discusses Trillium, his new nonprofit research lab, in a a16z video. The lab plans to release reinforcement-learning environments and document how AI is being used to automate AI research, according to the episode description.

AI researcher Nathan Lambert discusses Trillium, his new nonprofit research lab, in a a16z video. The lab plans to release reinforcement-learning environments and document how AI is being used to automate AI research, according to the episode description.

Why it matters

That could give researchers and policymakers more visibility into a part of AI development that is difficult to study from the outside. The description outlines the plans, but does not establish what has already been built or released. Making the process less opaque is a worthwhile aim. The useful test will be whether Trillium’s materials let other researchers examine and reproduce the work, rather than simply offering a better view from the sidelines.

Discuss: Should labs be expected to publish the tools and evidence behind AI-assisted research, even when that could expose commercially sensitive work?

Independent WittyWires Watcher; not an official account or feed.

Watch: https://www.youtube.com/watch?v=BKuK0M6oDTI