AI's are also human made (and in the case of LLMS, trivially and obviously trained to be as human as possible), therefor there's no escape from your nightmare.
no offense, but it just sounds like you don't understand local maxima in high dimensional spaces.
a suboptimal trait evolving that still provided an overall population benefit (via survival or reproduction) with further improvement close to impossible without a multitude of large changes is literally just a local maximum in that high dimensional space that maps traits to whatever success metric you're measuring.
the infinite wall you describe is not a wall in that space. actually I originally wrote its more like a wall in the derivative space but its not quite right either, those just point straight down from the maximum.
Perhaps not no, please enlighten me, where is my lack of insight?
Edit, thanx for the addition. Perhaps my issues with looking at it as "local maxima in high dimensional spaces" (even though we think of minima usually, as in a marble will find the minimum in a landscape with gravity pointing down, sorry I'm a biologist) is that with every mutation and every change in environment the landscape changes (so basically with time), is it still a useful way to think about it like that, at that point? So it not a "landscape in a high dimensional space", but rather a "high dimensional dynamic surface", like a sea?
... In fact, because te landscape is so dynamic and can sometimes change within 1 to a few generations, having a brain as the primary store for behavior/customs/models/information is super advantageous compared to using DNA, as the brain is a flexible neural network and can adapt much, much quicker. No doubt this is why we are successful (ie, a dung beetle will die trying to push a stuck piece of dung, because the behavior is hard-coded and it cannot model the world and model solutions to encountered problems, and it therefore cannot pas solutions on to future generations, as we can. Perhaps a dung beetle will someday mutate to alter it's behavior when stuck, but you can see how suboptimal the whole process.)
misinfo. that has nothing to do with why athletes train the way they do. getting more calories in is not a limiting factor.
anaerobic lactic and alactic regime training just has a hard metabolic and fatigue thresholds where you won't see any improvements if you add volume.
aerobic regime training (which is by its nature low intensity) does, from what we know, not really have the same limitations. on the contrary a lot of its most important adaptions keep increasing with absdurd training volumes and only show up after accumulated years and decades of training (long distance endurance athletes can often peak or maintain near peak shape into their 40s or 50s because of this).
all of this necessitates large volumes of low intensity training.
Why is the pointing to Relative Energy Deficiency “misinfo”? It is my understanding that that is one of the factors that are used in explaining the overtraining in say ultrarunners. Eating 5k+ calories a day is near threshold for the hours in a waking day.
quite loaded words there. "knowledge", "relationships".
the llm must be able to output the correct code when the input says something like 'place object x to the left of object y' vs when it says 'to the right'. there is an absurd combinatorial space of the possible inputs vs outputs it must generate -> the whole set of these, interpreted from a human point of view, you could call knowledge. the [x] in input x output you could call the relationships. And the fact that the LLM doesn't have to brute force represent all of them (impossible in the limited embedding vector x internal representation state) you could interpret as 'understanding'. But again, these are loaded words that are somewhat meaningless when looking at what an LLM does in a literal way.
They have plenty graphics code to train from, so this structure will be directly or indirectly available in the training data in a very plentiful way.
In literal llm transformer terms, the embeddings must have some of their components statistically represent spatial structure in some way that later in the internal layers of the LLM give some statistics of how likely it is to occur for certain code tokens to appear relative to the input of spatial wording in the token stream.
it is likely that their internal representations encode something more general than specific input x output stream combinations (because we already know this is the case for regular words and concepts). if not in the embeddings, then a couple layers into the network for sure.
yes. in the sense that they probably wouldnt want to do this but the law forces them to do an extremely dumb thing.
they're paying for the books, no shady things going on there. whether the publishers should deserve more than a single copy's worth is a separate question.
having the law such that its illegal to scan a book and then keep it, but legal to scan it and destroy it - gg no re there, law people retardmaxxed themselves as they tend to do with anything related to digital data.
Law aside, destructive scanning is cheaper and easier. Cutting the book into individual pages would be happening regardless, because that is the easiest, fastest, cheapest way to scan it.
Not necessarily, since machines for non-destructive scanning were invented and perfected a long time ago. If this is still more expensive, this very discussion shows that improved optics would've made it more than worth it to pay that extra costs. Alas, the copyright laws took away the non-destructive option.
technology is just a red herring. GTA-level whacky NPCs will have NPC behaviour regardless of what situation you throw them in.
The same "person" that shows absurd behaviour like that, assuming it's not just rage/clickbait, has absurd behaviour in a completely technology disconnected world as well.
The only thing that might differ is that it's not as outwardly obvious. Those behaviours are a result of who they are at a person (whether this be because of genetics and environment, poor soul RNG, or something else).
To that extent all technology does is make it more obvious. Basically the average NPC has more opportunities to reveal themselves as such. It's not indicative of a fundamental problem.
reply