simplyluke

5 ms·

31.
by simplyluke 4mo ago
Yes but many of the challenges directly contradict the idea that "coding is a solved problem"
32.
by simplyluke 4mo ago
I'm not really sure how this is exactly pertinent to a discussion on books being banned. A copyright holder can choose to cease publication for any reason, that's categorically different than a ban. Perhaps it's a point in a
33.
by simplyluke 4mo ago
Zero books are banned by name in the USA. Certain content is: Classified documents (although this is just illegal to share as the one with the original clearance, not to publish/read/possess after), child abuse material, and cop
34.
by simplyluke 4mo ago
> I think the technical part went fine but the interviewer was clearly frustrated by my low regard for LLM safety. I didn’t get the role. Anecdotally I've heard this is weighted as much as the technical interviews.
35.
by simplyluke 4mo ago
And yet you're still labeled a conspiracy theorist if you suggest that their might be ulterior motives for identity verification in the nation that arrested 12,000 people for what they posted online in 2023. I'm actually very su
36.
by simplyluke 4mo ago
I think Dario's about to learn very quickly that making overly-hyped statements around safety to generate market interest is likely to result in regulatory reactions that don't just entrench his position as a market leader, but al
37.
by simplyluke 4mo ago
That's why Dario's advocating for making open weight models illegal and also saying we should stop the clock on model development amongst the large labs.
38.
by simplyluke 4mo ago
Dario's been beating the regulatory capture drum for several years at various intervals, always in the name of safety, but it's hard to not see how self-serving it is. I'm personally very tired of reading the linear-algebra-m
39.
by simplyluke 4mo ago
Uber's comp packages are probably right in line with that. Tech salaries are trimodal, and uber's right in line with the big public tech companies. https://newsletter.pragmaticengineer.com/p/trimodal
40.
by simplyluke 4mo ago
> We will soon have to brace ourselves for an absolute draconian level of tracking I think framing it as inevitable is one of the biggest lies in the current AI narrative. No, I think the companies that are likely to be successful in the
41.
by simplyluke 4mo ago
Objectively I've always viewed this as the right approach, even just from an organization perspective. It's been something I haven't really adhered to in practice due to employers being pretty friendly with their tracking&#x2
42.
by simplyluke 4mo ago
Management bloat is definitely a thing in many companies, but there's a right level. Some number of C-level execs seem to think AI magically means the optimal structure of any company is an unlimited number of thousands of individual c
43.
by simplyluke 4mo ago
It's hard for me to not notice that the new C-level marching orders this year are that "measurement" jobs is actually what AI is killing (managers, HR, data, etc), and that seems to be an about face from IC-work being dead
44.
by simplyluke 4mo ago
> repair is expensive ($1000+ for performance cars, not that much cheaper for Civics) I can almost guarantee the lion's share of this is shop labor, not parts, and pressing a bushing out and back in would be more expensive after
45.
by simplyluke 4mo ago
I'm not all that sympathetic to small businesses that exist functionally as drop shippers for the same products with the same absence of support. Much in the same way I roll my eyes and go to 7/11 over the cute "local" m
46.
by simplyluke 4mo ago
The other thing that's changing is more and more CFOs are looking at the AI spend in engineering departments and hitting the brakes. Token leaderboards were cool when the spend wasn't a double-digit-percent of the entire departmen
47.
by simplyluke 4mo ago
> I'd pay a premium for even just a model that's 20% better The point I'm making is that I think we're rapidly hitting levels where corporate buyers aren't willing to pay multiple-times-more for marginal gains, a
48.
by simplyluke 4mo ago
My experience over the past decade has been being subsequently burned by being reliant on one provider's ecosystem after another. This is great until Reolink starts doing something shady to pad the bottom line and then it's on to
49.
by simplyluke 4mo ago
This doesn't break it down by experience, and I can't find specific data on that, but the recent spike in demand for engineers + subsequent drop in unemployment this year is well documented [1]. The demand for senior+ engineers ha
50.
by simplyluke 5mo ago
I do too, but I think it currently has a lot more to do with the quasi-recession we've been in since the end of ZIRP and AI is a better excuse to stop training juniors than telling investors it's belt tightening, just like layoffs
51.
by simplyluke 5mo ago
This is totally valid and I don't agree with the downvotes you're getting. Someone coming out with a 10x improvement is possible and would change the game immediately. The thing is, we really have been seeing marginal gains with
52.
by simplyluke 5mo ago
Maybe I misspoke by saying open source. The larger point I'm making is I think models are rapidly becoming commoditized. There is probably a small market long term that's willing to pay 10x for 10% marginal gains, but the majority
53.
by simplyluke 5mo ago
I'm already running a google TPU over USB on an otherwise very cheap board to do local computer vision on a front-door camera since I wanted to get away from Ring and other cloud services for that use case. And yeah, that may be the ~d
54.
by simplyluke 5mo ago
I think the bigger shift was harnesses and the two ended up somewhat commingled in people's minds. Claude code was a lot of people's introduction to using coding agents that could do a lot more than copy-pasting from a chatbot or
55.
by simplyluke 5mo ago
> that you can run locally That's doing a lot of work here. The future I see isn't most companies buying hundreds of thousands in hardware to run models, it's them adding a line item to their AWS bill. Inference costs on
56.
by simplyluke 5mo ago
Yeah, that's the part that just seems to be wildly under-discussed to me. If open source models are ~3-6 months behind SOTA, and ~opus4.6 capabilities are good-enough for product market fit, do the frontier labs have half a decade to
57.
by simplyluke 5mo ago
I mean, like any other skill that has pretty much been my experience (though I tried fusion + openscad), but there is something about being able to ask a computer all the dumb noob questions that makes that first phase easier.
58.
by simplyluke 5mo ago
Nor do I, but the loudest voices in public have spent the past 4 years telling anyone with a microphone that white collar work is dead. How would you expect that to make a new graduate feel?
59.
by simplyluke 5mo ago
I am (clearly) not as far down the rabbithole as the commenter you're replying to, but almost certainly not. Streaming 4k blueray is on the order or ~100Mb/s, which means on a LAN bog-standard gigabit ethernet and associated netwo
60.
by simplyluke 5mo ago
Yeah, CAD has been my personal example of "oh the barrier to entry for this skill was high enough that I didn't do it and now I can be passably bad at it enough to get some simple things done" I've had similar experience

More ›