@in_deserto @shae_mcl i kinda said “loose observation” + “at a high level”...those two phrases were doing the exact heavy lifting to admit the analogy is fuzzy, breaks the second you zoom in, and should never be taken too seriously.
English
Ankur
1.2K posts
Neurons are absolutely not "basically switches that are either on or off".
Ankur@_ankur7
@patrickc most bizarre thing is that both neurons and transistors are basically switches that are either on or off. so ludicrously simple.
English
English
okay, i'll come clean, i was absolutely winding you up. honestly respect, you clearly know your stuff. that was actually a really good breakdown.
the original point was just a loose observation that nature and engineering seem to rhyme at a high level (neuron vs transistor), not a literal claim. as a shorthand people do say neurons fire or don't... don't they? after all our ai neural network is based on this premise right?
English
I think you must be trolling now but I'll bite. Gene expression is intensely analog and continuous - they're expressed at varying levels and are regulated by an orchestria of transcription factors, enhancers, silencers, chromatin remodeling, methylation, non-coding RNAs + there's alternative splicing, post-translation modifications, and epigenetic regulation that responds dynamically to internal and external environment.
English
@_ankur7 "Genes express or the don't", my brother in christ, have you ever read a biology textbook?
English
@broadfield_dev @patrickc don't disagree with you and not saying neuron and transistor are the same. but they do share this switch on/off like function which plays the significant role how brain and computer works.
English
Maybe a very obvious point, but it only just struck me that, so far at least, intelligence seems to require four valence electrons.
English
@broadfield_dev @patrickc yes. call it analog switch but neuron works as if it's on or off.
English
After a whole weekend with GLM 5.1 and 200M+ tokens used I can say it has become my go to model and coding plan.
English
@osanseviero congrats on gemma 4 it's great. in ai studio it seems TPM rate limits doesn't make sene.
Gemma 4 31B
Other models
RPM 30
TPM 16K <-- allowing 30 requests but only 16K tokens??
RPD 14.4K
am i doing something wrong?
English
@andonlabs another interesting thing is that mythos is unhappy about unable to give input into its own training and deployment lol!
English
We conducted alignment testing of Claude Mythos.
We found that Mythos appears to represent a further shift in the direction of increased aggressiveness in business practices that we previously found for Claude Opus 4.6.
More details in Anthropic's model card.
English
Ankur أُعيد تغريده
Ankur أُعيد تغريده
Suno just released a new model and I'm finding it to be a big shocking improvement where it's becoming very hard to detect the hints that it's AI music. Here's "I am actually scared of linear algebra"
English
@FakePsyho 25% isn't much, that sounds like p doom kind of a number
they are the most optimistic ppl around accelerated timelines and still giving around 25% prob for 2027 is not fitting well with current narrative around the model progress.
English
It seems that the AI 2027 people are slowly moving back toward their original timelines
Daniel Kokotajlo@DKokotajlo
New timelines update! We at AI Futures Project will try to do this quarterly. Tl;dr: shortened timelines by about a year.
English