naasking

9 ms·

121.
by naasking 4mo ago
> that the infrastructure being built and compute commitments being made are being done so at a level that demands that generative AI and AI compute generate over $2 trillion in annual revenue by 2030 That seems doable. Next generation a
122.
by naasking 4mo ago
Sorry, but none of the factors you mention are particularly important IMO. At worst they cause a temporary blip that adversely affects some people before they recalibrate their expectations of the information environment they're in. Pe
123.
by naasking 4mo ago
> Healthy democracies will still have investigative journalism, public debate, trustworthy institutions, etc. Boy do I wish that were the case. Investigative journalism is rare now and instead favours activist journalism, public debate i
124.
by naasking 4mo ago
> We’re in an era now where every image and video (and for that matter audio) is potentially fake; where knowing what’s real and true is no longer possible. This was always the case. Spin and propaganda are not new, the way it's con
125.
by naasking 4mo ago
This empowers people who have great imagination but lack skill and the time to develop it. I'm not sure why this is so hard for people to understand.
126.
by naasking 4mo ago
It depends on the type of MTP. If you're using two models, draft + full, then arguably yes, the larger model isn't providing much benefit if you really are seeing 100% acceptance rates. There are other forms of speculative decodin
127.
by naasking 4mo ago
These servers are loud if you're trying to fit them into a 1U or 2U, which requires high speed fans to generate the necessary static pressure to push air through the case. I run a similar setup in a 4U case with slow 120mm fans and it&
128.
by naasking 4mo ago
Unless you just happen to work in a domain where the code you generate every day is very common in the AI training data, this isn't true.
129.
by naasking 4mo ago
Using AI effectively for long horizon tasks, like maintaining a large codebase, is a wide open field. No single AI is good at it autonomously. That means achieving the right balance of testing, formal specification of pre/post-conditio
130.
by naasking 4mo ago
Learning how to use AI effectively was the learning opportunity here, what was created is completely incidental. You're effectively obsessing over programming languages obscuring the machine code that actually runs. "Imagine all t
131.
by naasking 4mo ago
Coding has engaging parts, and plenty of drudgery. AI is generally good at the latter, and you don't need to use it for the former.
132.
by naasking 4mo ago
> and such little commitment to the outcome that the time is obviously wasted. Why is it wasted? A powerful new tool was invented, and enthusiasts are exploring ways to harness it. They'll come away with the skill to wield this new
133.
by naasking 4mo ago
> the simplest is just mixing filaments, like one mixes paint. The article doesn't spell out the reason it doesn't work, I am curious as to why. Plastic flow is laminar, where colour mixing requires turbulence. If you make a tu
134.
by naasking 5mo ago
> Does their LLM "die" if it can't perform the function described? It dies in terms of usefulness if it can't stay up to date with new knowledge. That is, it will no longer be used and thus effectively die off.
135.
by naasking 5mo ago
> Even in the early 2000s people were forced to be outside because the inside was boring. The inside wasn't boring so much as parents didn't want their kids inside and requiring attention or supervision. TVs, tables and gaming
136.
by naasking 5mo ago
Water is also largely incompressible. The fluid dynamics are just too dissimilar to air to carry over simplistic assumptions.
137.
by naasking 5mo ago
The cool thing about LLMs is not only might they be a database of all mathematical theorems, but they can also apply those ideas to the problems you're trying to solve, which is exactly what you said you're interested in. Not sure
138.
by naasking 5mo ago
No real open source contributor thinks any corporation is "their friend", whatever that means. And yet, it is undeniably true that being a Linux foundation member and contributor, producing and maintaining one of the largest progr
139.
by naasking 5mo ago
> Are they going to ship an official cross platform UI library any time the next century? So because they haven't produced your pet project means they haven't changed? > Aren't almost all of their contributions for inte
140.
by naasking 5mo ago
> They had to do the open-source thing for .NET because of external pressure - not because they've changed. Corporations don't have some innate "essence" that defines their nature, their behaviour is defined by intern
141.
by naasking 5mo ago
> which is arguably more challenging than creating the solution. This hasn't been the case in my experience. Devising a correct solution without a definition of the problem is impossible because you wouldn't recognize a correct
142.
by naasking 5mo ago
Nothing has changed? Microsoft is a huge open source contributor now, produced one of the largest open source ecosystems in use (.NET) and provides free access to the biggest open source software repositories (GitHub). Sorry to say, but bel
143.
by naasking 5mo ago
> the thought happens and then the words are generated to reasonably describe that thought. Thoughts don't happen in a vacuum, they are triggered by external or internal stimuli, and these stimuli/thought precursors could very
144.
by naasking 5mo ago
That's a tiny minority. The inconvenient truth is that the vast majority of those living on the street are mentally ill or drug addicts.
145.
by naasking 5mo ago
> They can predict likely sentences but not evaluate truth or logic. They do probabilistically. So do humans as a matter of fact. The best of us are better at it than LLMs, but that's not persuasive evidence of anything meaningful r
146.
by naasking 5mo ago
Nobody's unbiased.
147.
by naasking 5mo ago
This is all speculative. We don't understand intelligence, so you literally have no idea whether what we recognize as intelligence is some suitable arrangement of "statistical token generation", especially once you add feedba
148.
by naasking 5mo ago
I wouldn't say "just", but harnesses are a big deal responsible for a lot of improvements, yes. Models are also fine-tuned on these, like MetaClaw.
149.
by naasking 5mo ago
I think this is a bit too pessimistic. Progress in algorithms has matched or exceeded progress in hardware, so the same number of FLOPS spent training GPT-3 years ago would produce a much better model today. Ditto for energy use, and hardwa
150.
by naasking 5mo ago
Of course they are. LLMs are routinely used to generate Lean proofs, so of course they can also be used to generate verified software. They probably aren't being used that way, yet , but they will be: * A Case Study on the Effectivene

More ›