I’m working on the front end of a transcription app. It’s a lot of moving buttons and labels around to get them where I want, but there’s no way I would ever be able to do it with a keyboard as fast as using an llm. Are we putting ourselves at a competitive disadvantage to big tech et. al. by being anti ai purists?
Hàorán Che in fourth century Jiankang Jin province said ‘all competition is against the self’. But he married his horse so I don’t know.
An AI wont protect you from my fists! Put em up, old chum! 🥊😆
I consider “don’t use LLMs to write code” and “don’t use AI to generate art” as two almost unrelated concerns and moral dilemmas.
90% of the code I use to hand type at work was fundamentally slop code written for slop purposes. In non-software terms, there’s few reasons for your average plumber to spend hours hand craft pipes instead of using the magic pipe box.
Whether or not some important detail with legal liabilities like credit card processing works logically and safely in your software is pretty much the same with LLMs as the risk of outsourcing. You could already pay randos on fiver to write you mystery code for you to ship without even looking at it. It’s just that the lazy negligent path is faster now. nothing about LLMs really stops you from safe software practices, in fact those are faster and more accessible with LLMs now, it’s just a matter of you yourself choosing to vibe slop code or engage in engineering principles.
If AI-assisted coding becomes the long-term norm, it will probably be accompanied by significantly different programming languages, workflows, and tools (as well as more-capable LLMs). The current batch of AI-generated slop will be regarded as a transitional mess unmaintainable by humans or AIs, and any practices picked up in this phase will have to be un-learned later on.
In some ways yes, in other ways no. Would you rather be fast at stuff or have a sense of judgement, morals, and a wealth of experience? Maybe it has us beat at literal transcription, but it can’t make good human-oriented decisions about UX and UI.
The obvious answer is looking at the wave of absolutely terrible and unusable UIs, all of them pretty similar, and all of them vibe coded. They’re the polar opposite of interfaces I’d consider good. Yeah, the slop machine can slop them faster than a real programmer can make a good interface, but why bother if they’re gonna be bad.
Are we putting ourselves at a competitive disadvantage to big tech et. al. by being anti ai purists?
It doesn’t have to be either/or.
You can use LLMs as a tool and never touch Big Tech. You can use them for utilitarian purposes, without getting all cultish about them. You can actively hate “AI” while doing this.
Honestly, I think Big Tech is stoking the “Pro AI” and “Anti AI” divide, because what they’re most scared of is the public viewing it as yet another dumb automation tool. That would obliterate their business. They want the public to see it as an “all-in-with-us-or-nothing” affair, kinda like they did with social media.
Front ends are less about coding than about implementing something and then going through many iterations of looking at it and tweaking it. I don’t think AI can do the judgment part very well right now, but I’m not current on such things.
The Fediverse militantly against AI, you’ll get heavily skewed results on here.
Not using AI or at least not keeping up with how you can use it, you risk falling behind.
I can do in a day what would have taken me a week, and at higher quality using a powerful LLM.
Good comment, i agree with you
I mean yes and no. LLMs are a tool and you should be well versed in using them as a part of your toolbox. Don’t use it to replace proper planning or actually thinking about what you want to make. Also remember that LLMs are massively subsidised at the moment, figuring out how to run stuff locally and use less powerful models for stuff is a good idea. When the music stops llm tokens will drastically increase in price.
