naasking

48 ms·

841.
by naasking 1y ago
Car software is, perhaps counterintuitively, doing a lot more than the Apollo software did. Just think about the computers available at the time and how much memory they didn't have.
842.
by naasking 1y ago
> PLA is not strong and doesn't like to warm up in the sun "Strong" is ambiguous. PLA is stiffer than many other thermoplastics (resists bending loads well), but it's not as tough (impact strength is poor/it&#x27
843.
by naasking 1y ago
> the behavior of LLMs is not thanks to a learned world model, but to brute force memorization of incomprehensibly abstract rules governing the behavior of symbols, i.e. a model of syntax. I think reinforcing this distinction between syn
844.
by naasking 1y ago
Hossenfelder actually did a video on this just yesterday: https://www.youtube.com/watch?v=7See8OhtN-k
845.
by naasking 1y ago
> However, if you do the painstaking work of combing through someone else’s published work, and discovering that they are riddled with serious errors, including hundreds of incorrect predictions, you can post a pre-print to bioRxiv that
846.
by naasking 1y ago
Most realistic video games don't simulate all of the physics required. Even if we just stick to simulating the motion of the robot itself in an empty space, all of that data can be generated synthetically once at the appropriate precis
847.
by naasking 1y ago
A kinematic model of the robot is a physics simulation of the robot. I don't see why that wouldn't resemble real world data enough.
848.
by naasking 1y ago
Humans do not start as blank models, they have billions of years of pretraining from evolution.
849.
by naasking 1y ago
That's vision only mind you, no consideration of sound, taste, touch, and interoception. And GPT-4 is far more fluent and knowledgeable than a 6-7 year old where the total token count matches by this assessment (though I'm skeptic
850.
by naasking 1y ago
> Robots take in sensory data and return movements, in real-time. There is no large data corpus of this pairing to do self-supervised learning on, like there is for text. This is easily generated synthetically from a kinematic model, at
851.
by naasking 1y ago
Interpreting "just around the corner" as "this year" sounds like your error. Most projections are are years out, at least.
852.
by naasking 1y ago
"Can't" is a pretty strong word, effectively entailing "never". Never is a long time to believe computers can't think.
853.
by naasking 1y ago
I think you'll need something like Meta's Large Concept Models to get past the language and token barrier.
854.
by naasking 1y ago
> whereas someone from Spain is supposed to be Hispanic, which makes zero sense from any point of view. Hispanics are Caucasian in the original classification: https://en.wikipedia.org/wiki/Caucasian_race
855.
by naasking 1y ago
> Overall I would love it if medical research papers moved away from "race" and started getting more into the fine-grained genetic details. Regardless of the politics involved, this will lead to better medical treatments for ev
856.
by naasking 1y ago
> We are quite far into the development cycle of LLMs. No, we've barely scratched the surface. Billions of dollars have been poured into the stupidest possible thing that could work + scaling, and we're only now trying more cle
857.
by naasking 1y ago
How are you so convinced that no one is experiencing these problems?
858.
by naasking 1y ago
> But English is a subjective and fuzzy language, and the AI typically can't intuit the more subtle points of what you need. I disagree on the "can't". LLMs seem no better or worse than humans at making assumptions wh
859.
by naasking 1y ago
You mean assuming technology improves, which it has for centuries and given there is considerable incentive for it to improve, is naive? Seems inevitable. Do you think humans have achieved peak intelligence? If so, why, and if not, then why
860.
by naasking 1y ago
> Currently, the #1 spot sits at a score of 0.09, not 0.9. A far cry from being useful. The best intellisense and code completion tools would solve 0.00. Those were the only tools we were using just a couple of years ago. 0.09 is a treme
861.
by naasking 1y ago
> Most of what we write is not highly creative, but it is load-bearing, and it's full of choices. The idea that you can't specify the load bearing pillars of your structure to the AI, or that it couldn't figure them out by
862.
by naasking 1y ago
Great news if true, but I'm not holding my breath. Just adding units support doesn't quite go far enough, but C++'s backwards compatibility prevents it from enforcing the stronger requirement I mentioned.
863.
by naasking 1y ago
> Also, why keeping a history of all memory changes will prevent losing heat? How much power does a persistent storage (hard drive, SSD) require to preserve its stored data? Zero, which is why it emits zero heat. > Who needs to go bac
864.
by naasking 1y ago
> Your last point is what I was trying to get at — eg, reasoning just about your math expressions. I think it's borderline criminal that programming languages haven't added units of measure to their type systems yet, and then r
865.
by naasking 1y ago
I'm not sure these people aren't capable of understanding, I think it's more a product of what you said: their projects were also short and urgent, and that sort of environment is not conducive to understanding, particularly
866.
by naasking 1y ago
> Excluding hate speech and civil disobedience? Why would you exclude civil disobedience, one of the primary means by which you protest government?
867.
by naasking 1y ago
Lots of engineering disciplines work this way. For instance, materials science is still crude, we don't have perfect theories for why some materials have the properties they do (like concrete or superconductors), we simply quantify wha
868.
by naasking 1y ago
> Organizing things with made-up HTML tags because the models seem to like that for some reason. Where does engineering come into it? You just described one critical aspect of engineering: discovering a property of a system and feeding t
869.
by naasking 1y ago
> I suspect in most cases, people don't handle errors where they are thrown, but rather a couple of layers up from that Agreed, that's the point I was trying to make.
870.
by naasking 1y ago
This is the case anytime errors are propagated to callers instead of handled locally, which is probably most cases.

More ›