I don't think any of this makes much sense. The problem that this article ignores is that people build different things with AI, not the same things but faster. They build more superficial things, more throwaway things, things that are tailor-made for chasing the buck, and it just doesn't feel genuine or interesting.

It was much the same with search. Google search allowed us to find things faster at first...but over time websites responded to that and started doing SEO crap. Now almost every search results in a billion websites that say the same thing, but shallowly. Yes you can still find some gems but was it worth it?

AI makes things faster, but now it means we can target superificial needs very quickly...and so a new breed of crap software will be built with it. As long as there's a dollar sign at the end of the road.

▲

tptacek 5 minutes ago | parent | next [-]

I don't understand this argument. Why do you think everyone is "chasing the buck"? That describes 0% of the work I've done over the last 6 months. There's no "bucks" in the music players, macro trackers, and source code review tools I've built.

▲

doginasuit 2 hours ago | parent | prev [-]

AI makes things faster, and it pushes the envelope on what can be built. Like a telescope lets the human eye see further into space, an LLM helps track more ideas and structures within working memory. Yes, people will use this new potential to build slop. And other people will use it to build what has never been built before. The only question is which one we will prefer.

▲

aselimov3 2 hours ago | parent [-]

By definition an LLM is unable to build what has never been built before because you are going outside the realm of pretraining/RL. You might be able to combine components that already exist in a unique way, but even then you will be fighting against the mediocre making effect of LLMs

▲

outworlder 2 hours ago | parent [-]

That understates what LLMs are capable of doing. They can and do build things outside of their training dataset. As long as you can get there by following existing concepts, they can do it.

	▲	fwlr 28 minutes ago \| parent [-]
		Perhaps outside their training set in a strict sense of “this thing itself was not specifically in the training set” (although only perhaps - users often iteratively coax frontier models into achieving things just beyond their envelope, and those achievements could well be folded into the next model’s training set without you or I being aware of that). But even then, it is not directionally outside the training set: it steps outside not in the direction of the unknown, but in the direction of another part of its training set. An evocative metaphor would be dashing from cover to cover in the rain, because it lacks the umbrella or raincoat or simply the fortitude to tolerate getting wet that it would need to reach an uncovered destination.