SnoopJfalsehoods youtubers believe about "AI"
ben.biddington (NZ)hearing gullible 20-somethings say "this technology DOES have good use-cases, like in medicine for example…"
is going to turn me into the fucking Joker
Andrew Golding@SnoopJ I listened to the episode of This Machine Kills with Bruce Schneier and whoa on some of the claims Bruce mad there, same sentiment.
@huronbikes no kidding? That's a shame, I really like his writing
@SnoopJ they weren't the worst I've heard and he did bring up the big security problem in that data and instruction cannot be separated in a prompt, but eh, kind of some out-there claims.
_NetNomad
earbs 💖💛💙
Olivia Vespera@SnoopJ Does the protein folding one count as not a good use case?
@OliviaVespera I am not really "into" protein folding enough to hold opinions about that space, or tools that aim for it (e.g. AlphaFold)
I'm open to tools that accelerate the search over protein structures, especially since "it either conforms or it doesn't" and other clear criteria for goodness apply.
All the ones I've ever heard anybody say good things about predate the current craze for "AI" and I think it would be generous to assume this is what people mean when they talk about applications in medicine, but maybe some of them do think of this.
@OliviaVespera perhaps more importantly: the potential harms are much smaller than the "machine that diagnoses you" stuff, which is basically *made* out of harm.
@SnoopJ I agree, I wanted to be sure in case there is something I've missed. I've touted along with veritasium that this was probably the only genuinely goodd thing that generative AI has done. Not only to discover the shape of all proteins known to us, but to use that knowledge to generate new never before seen proteins.
@OliviaVespera the "generate new stuff" thing I have a lot more skepticism for
especially after google's stunt with GNoME which turned out to basically be a sort of advanced academic spamming of the materials community
David Fleetwood - RG Admin@OliviaVespera @SnoopJ I mean it's just as prone to inventing new proteins and wasting researchers time trying to replicate. There is no use case that does not face all the challenges the common uses do, it's a question of whether or not it can be quicker and less wasteful of time/energy/environment than historic methods.
I'd like to see it compared to say, Folding@home for instance.
davidak@OliviaVespera it is a good use of machine learning
the discussion is about generative AI like Large Language Models (LLM) or audio, image and video generators that are trained on copyrighted material
M Schommer@davidak @OliviaVespera @SnoopJ
Does anyone know of a good list that differentiates between the multiple types of "AI" technologies? LLM vs. MoE vs. …?
That would be helpful in such discussions...
Does anyone know of a good list that differentiates between the multiple types of "AI" technologies? LLM vs. MoE vs. …?
That would be helpful in such discussions...
MarjorieR@musevg @davidak @OliviaVespera @SnoopJ to me the main distinction between LLMs (Chat bots) and other forms of AI (machine learning) is that in machine learning there is some objective training data to be trained on (and with a non training set of data to validated against) whereas all an LLM does is predict the most likely set of words in a sequence based on its training set of words.
So in a medical context the model is tested and compared against clinical data or human expert judgement. An LLM applied to a medical question just comes up with plausible sentences.
So in a medical context the model is tested and compared against clinical data or human expert judgement. An LLM applied to a medical question just comes up with plausible sentences.
Orb 2069@musevg you would have to go pretty deep into the topic to understand all the different approaches
Wikipedia has probably a good neutral overview
https://en.wikipedia.org/wiki/Artificial_intelligence
and here LLMs specifically
https://en.wikipedia.org/wiki/Large_language_model
it is a big research field going back to 1943
i think the technology and research is not the problem, but for-profit companies acting unethical by ignoring copyright and making billions from the stolen work of (small) artists
@davidak @OliviaVespera @SnoopJ
Back in the day, we had expert systems, PROLOG, LISP.
Now I'm just looking for the right terms to discern the different types of "AI"… I guess, image creators like Midjourney and music generators like Suno aren't LLMs. Or are they? And the correct term for "AI" used to analyze images (medical or geo/spatial) is...?
@musevg LLM is for text and the big LLMs are multimodal, which means they also can work with image and audio, like you can speak to it, show something with your webcam and it can understand you and recognize it and answer you with voice, way more natural than Text to Speech (like OpenAI ChatGPT Voice Mode)
https://chatgpt.com/features/voice
those models predict one token after another
the models to generate media from text are diffusion models.
but there are now language diffusion models.....
@musevg i think usable terms to separate those two areas is Generative AI vs Machine Learning
where GenAI is the stealing slop machine and ML is what scientists do or pattern recognition in products like OCR
landel
Bill, organizer of stuff@OliviaVespera @SnoopJ Protein folding does not require (and most programs do not use) LLM-related tech. That form of machine learning long predates "AI" and has far fewer problems.
Robert J. Berger@SnoopJ This looks like an interesting take on it all. I've only watched this preview. https://www.youtube.com/watch?v=xkPbV3IRe4Y
