Post

all AI code i've seen has been total garbage they don't know how to refactor and simplify they do quite the opposite

It'll slow down when companies decide they're done making a loss on every customer and nobody can afford the bills any more.

economically speaking, it'll happen when investors finally realize they've been shoveling collectively billions into a furnace for several years now with zero hope of ever turning a profit

There's a lot of unethical practices in business. It doesn't mean we should all embrace them or use them in our own projects.

Oh, I grasp it. But I come back to the same question: how much of this is in pursuit of anything worthwhile? Do you look at the output of your average business/software company today and think "yes, this is useful". It's just speed, speed, speed in pursuit of... I don't know, and nor do they.

So many people reductively say "AI accomplished [feat]" without qualifying it in any way. It ported [x] game, but it's unplayable after a certain point, is composed of completely unreadable/unoptimized code, can only be updated by LLMs, and is undoubtedly comprised of stolen code.

They could, but they probably won’t. AI tools will allow many previously impractical projects to happen Also Rygar is mostly just an example of my testing over time to see how things change (been asking AI to port Rygar for a few years)

And again more broadly you're talking about massive amounts of air pollution, greenhouse gas emissions, noise concerns, energy usage problems, component costs, etc. all going up, and every time someone runs one of these programs it makes the boondoggle seem just a tad more worth it.

That would be the same dispute in which the original blog post from Alpöge and Buckmaster explicitly enumerated the LLMs they used in the process, yes cims.nyu.edu/~tristanb/st... The question in dispute there is “did OAI illicitly access A&B’s work for their own gain”, not “can LLMs do maths”

proofsandprompts.com and the guest posts on Terry Tao’s blog showcase a wide range of opinions, from “I for one welcome our new AI overlords” to “shun anyone who collaborates with the thinking machines”, but “actually they’re *not* good at maths” is not one you will find among them

Agreed. It will get you a result that if outside your expertise will SEEM good. It can build you a 3D model but if you understand topology you'll see just how bad that model is. It can generate a video but if you're a professional editor you'll see the continuity issues. And at WILD ecological cost.

That is no longer true. Certainly not for software engineering, although it was true until surprisingly recently.

Not a fair or direct comparison, but I draw parallels to VR, tech that I do like quite a bit. Can do 5 or 6 game genres extremely well and transformatively. What about social networking? No. Video conferencing? No. Mobile computing? No. But just making cool games isn't enough to justify the cost.