niplav @niplav@schelling.pt

PSA: numpy.isclose is neither symmetric nor transitive 💔

@gigabecquerel Dangerous to the world or to your wallet?

@Paradox Thanks :)

As it is said[1]

[1]: https://gwern.net/about#development

@erwinrossen The years of mindless data collection are starting to bear fruit 😃

👃

The Taoists were wrong:

Recent masturbation likely doesn't decrease meditation quality.

http://niplav.site/notes#Does_Recent_Masturbation_Decrease_Meditation_Quality

@cosmiccitizen Parmenides criminally undervalued

Understanding vNM oncoherence better -> prevent surrogate goals from being edited out of utility functions

Why is the majority of all living multicellular animals on earth nematodes?

IIRC ~half of all multicellular animals on earth are nematodes. What makes them special to be do successful?

It is the most important century

And I'm reading a report on snail welfare

Oh actually they make up >80% of all animals. (!!!)

@rq #gluehirne

@Paradox this assumes that the type of data learned on doesn't *really* matter, if it's video or sensory or text or whatevs

@Paradox maybe some self-criticism or self-reflection/chain-of-thought type stuff (constitutional AI in LLMs)

@Paradox yep, this very much moves towards brain emulation (from the top down or smth)

On WBE see this report which is forever on my reading list: https://www.fhi.ox.ac.uk/brain-emulation-roadmap-report.pdf

Not sure about the disanalogy to humans. I've heard people claim that humans learn suprisingly similarly to current LLMs:
* vast amounts of self-supervised learning (prediction of text in LLMs and of sensory data in humans)
* some reinforcement learning on top (action-reaction in humans and RLHF in LLMs)