261–270 of 283 posts
Any engineer (any person actually) can “learn to use AI” in a couple of days. It’s not rocket science; there’s no chance of left behind. If you haven’t use LLMs at all, a weekend would be enough to be on par with everyone else in the industry
Just learn, sure. But the difference between my efficiency of using it on my day 2 and month 6 is significant. Yet I feel I am barely scratching the surface of it.
Earlier quoted context omitted.
> I've come up with a set of rules that describe our reactions to technologies: Anything that is in the world when you're born is normal and ordinary and is just a natural part of the way the world works. Anything that's invented between when you're fifteen and thirty-five is new and exciting and revolutionary and you can probably get a career in it. Anything invented after you're thirty-five is against the natural o…
It's not so much the technologies that are good/bad, but IMHO that the pipeline between invention of and abuse of technology is increasingly short.
Earlier quoted context omitted.
If the models get to a point of total consistency there's still a LOT that we need to figure out and learn about how to use them. Let's say models can exactly and correctly write any code you ask of them. - How do you break down a project into a sequence of requests to models? - How can you most effectively parallelize the work - models will never be instant, so there will always be benefits in working out how best t…
I don't mean to be dismissive, but aren't you describing simply working on projects with a finite amount of human hours/capital, but substituting finite amount of tokens? I recognize all of those problems as project problems long before LLMs, so I'm not sure it addresses GPs point.
You need to be able to estimate how long an approach is likely to take in order to correctly prioritize your work.
Coding agents throw a wrench in this because some of the stuff that used to take a long time doesn't take a long time any more. I'm finding my 25+ years of experience in estimating software has been completely thrown off - it's taking a whole lot of work for me to start building up those intuitions about what's quick and what takes a long time.
Earlier quoted context omitted.
With current models, you're right, there will be nothing to show for the effort except the code itself. I suspect that will change sooner or later. Models will be cultivated over time the way we cultivate full-time employees now, with an acquired awareness of what they're building, new skills picked up in the process, and insight into how the larger system works.
How do you reckon? There's a long-running instance of the model on the provider that's allocated to my organisation? Or are you thinking more of a server-side memory system, similar to the (currently very fallible) ones like Honcho and Mem0? What happens when my org stops paying that provider? Do I get to take the now senior agent with me to the next provider? Does the provider have to delete it (and all that learnin…
Karpathy's notion of a self-maintained wiki may suggest a direction for work in that area. As he points out ( https://gist.github.com/karpathy/442a6bf555914893e9891c11519... ) this is basically the latest of several attempts at implementing Vannevar Bush's "Memex" concept. I think it will find application in some form at organization-wide scope.
If the trend towards centralization holds, then yes, who gets to own and maintain the state in a stateful AI model is indeed going to be a big question. I hope it doesn't hold.
Earlier quoted context omitted.
I WANT to be excited by LLMs. But it has been so frustrating to listen to the hype and then contend with the actual results. How bizarre is this tool, that I have been forced to conduct interviews across all domains, to see where people are having actual productivity gains? I never had to do this with the iPhone, or with word, excel or any number of cool technological inventions. The early internet empowered people,…
I have been treating LLMs as a research assistant (and sometimes tutor) as I have explored analog computing for a hobbyist project that I have been working on for 6 months now. To my mind, saying the AI will cause me to forget how to learn is equivalent to suggesting that going to school for instruction will cause me to forget how to learn .
I see plenty of “classically trained” people (as opposed to self-taught) that have no idea how to learn things on their own and at their pace.
I mean, in the era of youtube and the internet, any subreddit is filled with posts like ‘where do I start with X?’ or a endless rehashing of the same beginner questions from people that are not even able to use Google.
I am biased as a self-taught person that dropped out of high school, and it is baffling to see people have forgot how to learn things by themselves past 10 years old. There wouldn’t be so many people with student debts if that were the case.
Any engineer (any person actually) can “learn to use AI” in a couple of days. It’s not rocket science; there’s no chance of left behind. If you haven’t use LLMs at all, a weekend would be enough to be on par with everyone else in the industry
Earlier quoted context omitted.
It's a bit arrogant and borderline Luddite to suggest that 'your era was legitimate' and that somehow these new things which you don't understand are somehow 'lesser' or illegitimate. In the long arc of history, I'm doubtful we'll see 'the last 50 years' as 'the Golden Age' - that's just a personal, contemporary romanticization. More than likely, the advent of computers -> web -> AI etc. will be one block of the 'inf…
There is no doubt that AI is world changing technology. I'm not sure I want to go back to the world before LLMs. However, he's right to lament the impending demise of personal computing . Our computing freedom is being attacked on all fronts by governments and trillion dollar corporations alike, and things are not looking good for us. Our machines are increasingly locked down by rent-seeking corporations. Software is…
Thanks to Chinese vendors, there's a ton of near Tier1 AI that we can all access.
Unlike Android where it's not feasible to use though it may be Open on some level.
Black-and-white thinking like this is not healthy. You can still do creative thinking while using AI as a powerful tool at your disposal. Some mathematicians like Terence Tao are comfortable doing this, for example.
"The reasonable man adapts himself to the world: the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." Richard Stallman was unreasonable. So was Linus Torvalds. I'm hoping that something wonderful and entirely human-centered will come out of the anti-AI movement, up to and including a bifurcation of the internet.
It will. That's why it's important to be a "Luddite" and to make our voices heard. Plenty are on the sidelines, plenty dislike AI, and plenty don't care to drink the kool aid and would rather exercise empathy towards fellow humans than welcome the era of the robotic overlords and their billionaire masters.
This place is an echo chamber and doesn't reflect the views of the vast majority.
Earlier quoted context omitted.
There is no doubt that AI is world changing technology. I'm not sure I want to go back to the world before LLMs. However, he's right to lament the impending demise of personal computing . Our computing freedom is being attacked on all fronts by governments and trillion dollar corporations alike, and things are not looking good for us. Our machines are increasingly locked down by rent-seeking corporations. Software is…
Those are all fine points, but 'lack of open compute' isn't quite an AI issue. Thanks to Chinese vendors, there's a ton of near Tier1 AI that we can all access. Unlike Android where it's not feasible to use though it may be Open on some level.