The rain game on March 6 was one of the wildest sporting events I've ever attended. Puddles stopping the ball and everything. #woso #uswnt

And that's highly related to LLM evaluation in general. When outputs are long strings / conversations, how do we compare to some ground truth dataset and produce a metric value?

San Diego was about 19.5k, and Angel City and Portland just over 18k. Per www.sportspromedia.com/news/nwsl-re...

@arsenalwfc.bsky.social Arsenal; England's FA Women's Super League

Huh. People tend to write down things that are notable. Corpuses are then the residuals of the human experience. So AI will by default produce things we least expect or are interestingly expressive (= art, hallucinations).

I was watching Brooklyn Nine Nine and reflecting on the absurdity and intensity of individual character choices and development. Hyper-expressive characters acting much more Extra than normal people would in the same situation with the same motivations. AI is learning humans-as-art in many ways.

If you'd like to learn more, our R&D program is learning-engineering-virtual-institute.org, our prize competition is tools-competition.org, and our listserv (just use the join request) is groups.google.com/g/learning-e... (scientists, technologists, funders, educators).

No, I was referring more to your comments about AI tutors in general, not Altman's huge claims. The learning science community generally isn't chasing the hype machine but is evaluating AI just like other education interventions. We really do care about rigorous research designs.

Learning scientists do care about this and the education research community is leading rigorous evaluation of tools that use advanced computation, from psychometrics to AI. See rori.ai/learning-out... for an early example.

That's pretty funny. Silicon Valley Noir

What is posting if not a writer's dream deferred

I recommend her thoughts at substack.com/home/post/p-...