jj4211

Microsoft already changed what they are selling. Now they are renting document storage that happens to have document modification software integrated.

They long decided that owning access to the data was more leverage than the software used to edit it. Ditto for Adobe. Pretty much immediately software as a service encouraged "don't worry about ever saving/managing your data" and it wasn't about user convenience...

You are right no one cares how it is implemented, and on select contexts, right that LLMs can produce sufficient implementation.

However it isn't generally the case that they can produce quality software. The bad news for a lot of developers is that by volume most software isn't exactly great quality either, particularly in house enterprise software is already crap, so there is large area where differently crapi isn't so bad.

I will say that people who want their software well supported are likely to be ill served by "vibed" solutions. It is really tricky to debug complex software that was vibed, whether human or GenAI. If it worked perfectly the first time around and/or the software in question is fairly disposable, well those may be fine.

Well, depends.

Software is subject to copyright. Yes you can implement the same APIs, but when you get into the visual presentation, things can get copyright related even if the code isn't being ripped off.

If you decompile rather than clone, well then the code is copyrighted even if transformed into plausible source code.

Of course then there are patents...

Possibilities include the current companies surviving, but after financial catastrophes, some falling completely, and getting absorbed or superseded by big players (e.g. if Anthropic ended up somehow financially vulnerable, Microsoft might slurp them up, or Microsoft decides to do their own initiatives in earnest).

Yeah, random stuff happens and normal people don't assume something like a pregnant mistress.

One time someone left a few roses under my windshield at work on valentine's day. I assume someone mistook my car for someone else's, not that I had some secret admirer who wanted to do a pointlessly anonymous gesture.

One time someone let out the air from one of my tires overnight in my driveway, for no apparent reason and it caused me mild annoyance.

Anyway, fake internet story for fake internet points.

This may be the reality, but in theory the answer is for people to say "fuck working for these asshats if we are stuck at the mercy of caprcious users tipping".

I would argue there is no ethical way to use them.

He didn't mean to question the premise that the consumer market is being starved, he was highlighting the reality of that and connecting it as being motivated by a fear of local models.

Admittedly, the question was ambiguous and could be taken either way, but the comment didn't mention that consumer market was starved, his question brought it into the discussion.

Note that I don't think that it is fear of local models personally, I think it's fear of each other. I think cloud infrastructure providers already proved that even when people can locally host stuff, they mostly don't. So many consumer electronics could meet their promises all locally, but still force the users to connect to their internet services. The tech market has already proven time and time again they are suckers for subscription eshitification, no need to panic over it being different for AI.

A bubble burst is not necessarily 'the end' of a tech. The internet bubble burst back in 2000, and here we are.

A lot of the financial burden is offloaded to SPVs that may well be allowed to fail, as the AI companies don't directly carry the liabilities. They certainly carry enough of them, but they have done a lot to manage to leave other entitites holding the bag. Government may happily let Anthropic fold if it came to it, because Trump is all pissy with them. SpaceX would depend on whether Trump is pissed or loving Elon in the moment of decision. Hell if Trump decided that 'TrumpAI' was a possible concept he'd probably let them all burn for the chance of his grift picking up the pieces.

Yeah, there are scenarios that it can 'one shot', either very straightforward variations of common themes or a use case that is comprehensively instrumented in a way to "know" if it is correct or not automatically, and having the solution iterate against that criteria evaluation until all criteria pass (though in the latter case, I have seen it change the criteria deciding the criteria was a mistake instead of fixing a problem).

If you spend enough tokens, let it rebuild and test on its own, and let it churn and churn, then I'm not too surprised it could make this launch and be superficially intact. The real pain starts when trying to close the gaps the project has in this sort of scenario.

I will say that GenAI makes golang a lot less tedious. Lot's of obvious but required boilerplate to force the developer to abide by consistent structure. Though I still prefer reading python code since I don't have to peruse a ton of boilerplate there.

Enjoy

At least my experience is that a vibe coded app gets to 'early alpha' then stays there. If AI is augmenting a developer who is otherwise competent and they don't let it get ahead of their own understanding, it can work out though not as 'miraculously fast' as this. If AI can oneshot an application then it can also work out. If a project vibes itself in a week or two to 'looks like what it should be but isn't good enough', then thus far every one I see is doomed to sit there, because the AI can't finish it and if you let the GenAI produce that much code, it's generally not serviceable even for very capable developers.

I spent a lot of time last week basically peeling back an overly vibed attempt at an application by a colleague. It was too flaky and incomplete and deleting large sections of it was frankly more practical than trying to rework it. It had gotten to the point where every attempt to try to have GenAI fix it one way would just break something else. That project was, essentially a fairly small command line utility, I fear to think what something the level of a Photoshop knockoff would be like.

Read a review by someone, this vibed up solution isn't going to win anyone over either because it evidently sucks. It may imitate the UI more closely for what works, but not much works adequately.