@littmath I'd probably throw an 'if the work is of an impersonal and output quality is the chief priority' in there if we want common sense to settle this

@littmath What about if at least one of the points of the output is to get better at making the output?

Obviously as capabilities improve some tasks move from the second bucket to the first, and exactly when that happens depends a lot on one's own distribution of skills, etc.

@littmath Worth mentioning because aside from spiritual concerns there's actually been some econ discourse lately about ideas like junior jobs subsidies, for ppl who are not yet better than a frontier LLM at a field but can plausibly train to be

@peligrietzer Sure, I guess that was implicit.

@pli_cachete Yeah I was trying to gesture at this kind of thing with "on balance." Of course there's some cost-benefit analysis to be done here.

@joseph_h_garvin Right, but if one's outputs are important you might prioritize them over personal development of some arguably less important skills. I agree there's a cost-benefit analysis to be done here though.

@littmath If something like this comes to pass then for ppl in these jobs using LLMs whenever it optimizes output might beat the purpose

@littmath I at least hope to improve the quality of my output instead of the quantity. So focus on creating good work myself and then expanding and refining it with AI. In my experience if I build the foundations with AI then I just lose control and learn nothing.

@littmath Also - even if you do acquire the skills, there are some activities too humdrum to bother over. Front-ends, fonts, colour schemes, LATeX formatting. I appreciate good output but can’t be bothered. Perfect use of AI.

@littmath I find that when I use it for a task I consider myself mediocre at (also coding), at first I think it’s accelerating my progress. But then I spend more time debugging what it did and learning nothing from the process. There’s a bigger lesson in here that im too frustrated to see.

@littmath Sometimes - e.g. creating a web page to demonstrate a topic it’s ok to let the AI design if you don’t have specific requirements. For other things, I find that at tasks you’re competent at you get a lot of value out of being able to distinguish important results from not

@littmath You should certainly use AI to write, regardless whether you are competent or not, if English is not your first language.

@littmath IMO for at least some applications one might want to accept some decrease in output quality for an increase in speed/productivity.

@MarkusBaumannAi Obviously fine to practice with the tools before they improve your outputs, but it's not clear to me why you'd delegate to them before this.

@littmath I think they're good for brainstorming. Like if you want a quick reminder of something you already know - and yeah LLMs for that is better than googling.

@littmath Careful with the balanced and sensible takes on social media, they'll eat you alive

@littmath This may prevent you from becoming competent though

@littmath @joseph_h_garvin I think sometimes people like myself can also overvalue code quality- there are scenarios where what the code does matters a lot more than how it does it.

@littmath Harms my outputs today is the wrong metric, these tools improve extremly fast, and the skill of using them compounds.