jcgrillo

28 ms·

571.
by jcgrillo 11mo ago
Did any users actually move to chatbots? By which I don't mean the 0.001% of tech nerds who buy chatgpt subscriptions, but in aggregate did a meaningful number of google searchers defect to chatgpt or other llm services? I really doubt
572.
by jcgrillo 11mo ago
> They consistently have the best or second best models. This is the problem with your original argument. It assumes that having a "good model" (e.g. one that performs well on some benchmarks) has somehow to do with something
573.
by jcgrillo 11mo ago
This is an excellent metaphor, so don't take this as criticism merely an observation, but it skews heavily towards the techno-utopian narrative that scam artists like Altman and Pichai keep harping on. Your techno-dystopia makes the sa
574.
by jcgrillo 11mo ago
In the article Pichai is quoted saying: > "It doesn't matter whether you want to be a teacher [or] a doctor. All those professions will be around, but the people who will do well in each of those professions are people who lea
575.
by jcgrillo 11mo ago
> frontier labs, which had already produced a product that was going to eviscerate Google Search (and therefore, Google ad revenue) > If Google does nothing, they lose. Is any of that actually true though? In retrospect, had google
576.
by jcgrillo 11mo ago
I've heard the same breathless parroting of the marketing hype at large O(thousands ppl) cloud tech companies. A quote from leadership: > This is existential. If we aren't early adopters of AI tools we will be left behind and
577.
by jcgrillo 1y ago
The next time you're working on your car google bolt torque specs and cross reference the shit their "AI" says with the factory shop manual. Hilarity ensues.
578.
by jcgrillo 1y ago
Indirectly related, for anyone interested in the topic, Pirault and Flint's Opposed Piston Engines[1] is a nice survey. Unfortunately it seems to be commanding a shocking price these days though. [1] https://www.amazon.com&#
579.
by jcgrillo 1y ago
There are people working on internal combustion engines with a very similar design currently, for many applications (military, trucking, etc) diesel or diesel electric is the only realistic option for the foreseeable future: https:/&#
580.
by jcgrillo 1y ago
Executable "toy models" are excellent in practice for property based regression tests. The invariant is that for all inputs the behavior of the oracle ("toy model") is the same as the behavior of the system under test. A
581.
by jcgrillo 1y ago
You touched on this but to expand on "numbers driven bullshit" a bit, it seems to me the biggest drag on true innovation is not quantifiability per se but instead how organizations react to e.g. having some quantifiable target.
582.
by jcgrillo 1y ago
Real world personal connections are how we establish trust. At some point you have to be able to trust the people you're collaborating with.
583.
by jcgrillo 1y ago
I guess the main question I'm left with after reading this is "what good is a prototype, then?" In a few of the companies I've worked at there was a quarterly or biannual ritual called "hack week" or "inno
584.
by jcgrillo 1y ago
> Some go so far as to say “AI not welcome here” find another project. This feels extremely counterproductive and fundamentally unenforceable to me. But it's trivially enforceable. Accept PRs from unverified contributors, look at
585.
by jcgrillo 1y ago
> For our team, every commit has an engineer's name attached to it, and that engineer ultimately needs to review and stand behind the code. Then they claim (and demonstrate with a picture of a commits/day chart) a team-wide 1
586.
by jcgrillo 1y ago
How would you handle production incidents in such a codebase? The primary focus of a software engineer is to make the codebase easy (or at least possible) to understand. To tame complexity while achieving some business objectives. If we&#x2
587.
by jcgrillo 1y ago
IME devs actually do precisely the opposite. They write code and then ask the LLM to do the "boring" part and write the tests for them.
588.
by jcgrillo 1y ago
Apollo made about a half dozen crewed trips to the moon (quantity deliberately left vague to account for quibbling over technicalities). They made it a relatively reliable, repeatable affair. My point is that organization--the team of peo
589.
by jcgrillo 1y ago
> Um yeah don't use liquid oxygen. You could probably do it once..
590.
by jcgrillo 1y ago
The goal could be simply to learn how to do it again, since almost everyone who actually has done it--on any level, be it engineering, management, manufacturing, flight crew, ground crew, etc--is dead. That's a totally worthwhile exerc
591.
by jcgrillo 1y ago
And most of the people who actually did it aren't alive anymore. A corollary from some other recent tech news: https://news.ycombinator.com/item?id=45649178
592.
by jcgrillo 1y ago
Apollo was much better value for money. It inspired generations to study and enter STEM fields, it gave us multitudes of technological advances, and it gave the entire world something to marvel at. It gave us the earthrise image, which fuel
593.
by jcgrillo 1y ago
Every time I set up a new machine--work, personal, whatever--the first thing I do is set up my rust toolchain and the second is 'cargo install ripgrep'. I really enjoyed your talk at the Boston Rust meetup a few years back on fini
594.
by jcgrillo 1y ago
What a beautiful lecture
595.
by jcgrillo 1y ago
You don't get to excuse a lapse in situational awareness because one or another of your senses is compromised. Compensate. Concentrate. Whatever it takes. Or, of you find that you're sufficiently impaired that you're not comp
596.
by jcgrillo 1y ago
Agreed, but the root cause of the issue the emergency vehicle operator are noticing isn't sound deadening or infotainment gizmos, it's inattentive distracted drivers. Probably playing with their cell phones instead of driving. Sim
597.
by jcgrillo 1y ago
If you have even the slightest amount of situational awareness sound deadening or infotainment gizmos will not render you unaware of an approaching emergency vehicle. Scan your mirrors, scan your instruments, scan the horizon, repeat. The p
598.
by jcgrillo 1y ago
As a happy owner of a 30 year old Toyota 80 series I'd recommend getting one! The computer can be diagnosed with a light bulb. Or go even older like a 40 series Land Cruiser. No computers anywhere in those.
599.
by jcgrillo 1y ago
If a power failure during the upgrade causes some unrecoverable problem that is a serious design failure. The answer isn't "make power failures less likely" instead it's "make the update process robust to power fail
600.
by jcgrillo 1y ago
Exactly. It has become accepted that manufacturers can sell us complicated systems before they're "done" and software is the excuse. It should not be acceptable, and if done well we could see incentives against this behavior

More ›