Half of that sounded suspiciously like AI slop. You might have an actual message, but you badly butchered it, especially by spamming diffusion. You seem to be using the language of statistics to explain the functions of the brain, why we err, failures of our society, and you are throwing together a lot of things at once, which is typical of say…an ADHD person, who sees it all as connected.
You badly structured it, and you sound like a scizophrenic, but I think I understand what you are trying to say. I certainly will not put up with that much to not become a pariah, I almost got into a physical altercation more than once with a co-worker for this reason.
Peanutbjelly@sopuli.xyz 2 weeks ago
will try to take it in good humour, but i love how i got compared to ai, adhd(AuDHD would be the real wombo combo here so you get points), and schizophrenic people.
and i would hope i don’t confabulate half as much as an LLM.
although an understanding of the modern situation does require an unfortunately theoretical take, while there’s more noise and conspiracy theories being socially reified than most people can remember. but i’d like to think i’m weighting this take via the best available expert consensus that i can find and source. biggest ‘correction’ i’d make is that i was beaten black and blue for waiting outside of the library, which was unrelated to the protest.
if you do actually care, and can handle more than the usual 140 character tweet limit, here’s some elaboration.
the ‘sycophancy into delusion effect’ i refer to can be seen widely reported on most news sites, where chatgpt and the like cause a feedback-loop into a psychotic break. predictive processing and the bayesian brain are leading theories in psychology that work well nested with other leading theories such as global workspace.
this is a very recent example with michael levin from tufts, who often works with friston’s free energy principle and active inference (included notes in wiki)
friston has hundreds of thousands of citations, if you care about pedigree. i hope i do not poorly capture or inaccurately represent any of their ideas, but if you’d like to drink from the source, you have my full recommendation.
that’s where the “saving energy” stuff comes from. while DKE might not perfectly and accurately explain the situation, i’m all for better ways to convey that eco-niche specific intelligence doesn’t always transfer, especially if it’s ‘overfit to a local minima.’ in the bio-world you have overspecialization, like pandas too fit to a specific environment, which may focus on skills that don’t transfer outside of that environment. there’s a lot more to gain from the full bayesian perspective, but there is a lot to be gained just by looking at how systems can successfully co-construct, and their possible failure states that are inevitable as systems grow apart into new niche environments.
there’s actually an interplay between that ‘energy saving’ property and putting energy back out which can be used to explore the environment, build a more robust model, and survive greater environmental shifts. this is explained in active inference. good, but slightly old textbook on MITpress.
i’m saying that meta-awareness of the failure states in these specific system dynamics could do much more general and robust good for society than being socially pressured into climbing the socio-economic hierarchy as hard as possible.
there’s a term for an imagined AI going rogue due to being overfit to a single goal. this is called a ‘paperclip maximizer.’ i compare the current socio-economic system to that failure. you know, ‘capitalism number go up!’
i don’t think any studies i’ve seen disagree with that take, but if there’s a relevant expert who’s got a strong weighting i’m unaware of, i’m always open to updating my weights.
as for learning yourself into some information bubble, or how someone can hold ridiculous beliefs without the need to question them, such as grand confidence despite low evidence, is often by taking something you have low evidence about, and having high confidence. and then giving it a high weighting. funny enough, friston’s dysconnection hypothesis is about framing schizophrenia as precision weighting issues, but i don’t think they are the kind i have TY.
mahault has a phd under friston, and her epistemic papers are essential IMO.
so there you have it, the larger environment of my thoughts, largely focused around one of the most cited neuroscience experts of all time, and michael levin who i mentioned is doing some of the coolest current empirical results in modern biology.
i tried, thank you if you got this far. if nothing else, please stay curious, but beware information silos that disable coms completely, or otherwise create barriers to properly comprehending the systems being represented. ‘nothing about us without us’ is important for a reason.
WorldsDumbestMan@lemmy.today 2 weeks ago
Sorry man, I’m employed.