#that you see in biological brains (unlike feedforward neural networks) | Explore Tumblr posts and blogs

positivelyapunovexponent-good · 2 months ago

Text

[/Image ID- screenshot of a tumblr reply reading as follows:

"AI" -> machine learning. and the winners in question do not own large language models like ChatGPT. they themselves warn about the responsibility of using machine learning ethically. i cannot believe that i have to say this but machine learning / "AI" has legitimate and critical uses for healthcare, scientific discovery, etc. common examples are AI used to ID cancer cells etc. you HAVE to start disentangling what we commonly call "AI" ie. the product that is sold to you the consumer and AI used in valid and vital scientific research ie. the tool. but then again this is the piss on the poor website. ffs

/End ID]

(Source)

#Fucking THANK YOU to all the reasonable people in the replies #I stg the reactions to the 2024 Nobel Prize in Physics on this hellsite have me chewing glass and NOT IN A GOOD WAY #Stop fucking shitting on Hopfield if you don’t understand what he did enough to actually make a nuanced critique of the decision #For my own part I think his work on associative content addressable networks is beautiful and fascinating and revolutionary #Like he laid the groundwork for studying network architectures that actually reflect some of the dynamics and associative memory structures #that you see in biological brains (unlike feedforward neural networks)#And the fact that you can model these processes with a simple cyclic graph- like that’s enough- is so beautiful #One of the things I learned doing my undergrad thesis is that cyclic graph neural network architectures can pack a powerful punch #Not in spite of but BECAUSE of their simplicity and symmetry. It genuinely shocked me at the time that was not what I was expecting #And none of this has to do with ChatGPT oh my fucking Gd not everything has to do with ChatGPT give it a REST #Anyways. Hopfield networks are a gorgeous beast and if I could spend the rest of my life researching them I’d be a very happy mathematician #The fallout from the prize decision makes me really sad because- as much as I’m already used to it from being a math person on this site-#it hurts seeing strangers get up on a soapbox and tear apart my special interest without even understanding it #just because the media described it using the wrong buzzword. Like genuinely hurts. The vitriol of it all…#Maybe I should make my own post describing exactly what I love about Hopfield networks #Maybe other people will see why they’re beautiful #That would be really nice #hopfield networks #machine learning #2024 nobel prize in physics

45K notes · View notes