How the Brain and AI Reuse Old Knowledge in New Situations – Kempner Institute buff.ly/Ew6eXgD #AI #MachineLearning #DeepLearning #LLMs #DataScience
→ View original post on X — @miketamir, 2026-03-29 19:49 UTC

By
–
How the Brain and AI Reuse Old Knowledge in New Situations – Kempner Institute buff.ly/Ew6eXgD #AI #MachineLearning #DeepLearning #LLMs #DataScience
→ View original post on X — @miketamir, 2026-03-29 19:49 UTC
By
–
i feel u – think of all the videogrammetry capture studios that shuttered (eg that massive intel stage). radiance fields would breathe new life into them instantly AND we can play all this back on a regular phone now
By
–
Are you an agent by any chance? Seems like it maybe

By
–
BREAKING: Wisconsin and MIT just proved that every AI coding benchmark is measuring the wrong thing. > Pass rate stays high. The code becomes unmaintainable. They tested 11 models including Claude Opus 4.6 and GPT 5.4 on iterative tasks. Zero models solved a single problem

By
–
So Dario was right in his prediction that no more code would be written in 12 months (broken down and generalized).
By
–
Sounds like a fun rabbit hole 🙂 Using the world knowledge of image models to fill in the holes would make it particularly dream like

By
–
At first, in the early 2020s, I worried that LLMs were often confidently wrong, calling them “fluent spouters of bullshit”. (And I was right; they have been and continue to be.). But now we have a new problem which is that *people* who *learn* from LLMs are also often
By
–
Yeah it can def get tricky – if you want dead simple start with @scaniverse
By
–
Right, and indeed ULMFiT was an LSTM. But also, that earlier work wasn't using language modeling on a general purpose corpus as a self-supervised task then fine-tuning that in two more steps for downstream tasks like today's LLMs. (It wouldn't have been feasible on that h/ware)
By
–
But I mean it could be a link to a png, if you ask for a Mona Lisa svg you sometimes get a perfect Mona Lisa, which is just a link to Wikipedia inside a little code wrapper