@hipsterelectron @fasterandworse 2) put together a small group of human collaborators, lay down the foundation for the thing you want to build, get a basic product out the door, and gradually build up your team by hiring more people and training them, and so on. Personally, I'm trying as hard as I can to think about this with only greedy intentions, and option 1 still does not look attractive.

@hipsterelectron @fasterandworse so, speaking purely from the perspective of let's rub together two moneys and see if we can make a third money, you could 1) use a tool that tries to makes your code more similar to code in its dragnet training corpus based on prompts, learns nothing, makes a mess of everything, and mostly creates messes for the bored human technician to clean up, or

@hipsterelectron @fasterandworse what is really great about human collaborators is you can explain things to them, which they can understand, when they don't understand they ask you for clarifications, and the experience is cumulative over time, so they get better at the thing the more you work together.

@hipsterelectron @fasterandworse in contrast, I ran a similar experiment the last two months where I hired a friend to help me with one of my side projects, and it was worth every penny. Working with a skilled human collaborator for two months shot the project ahead way faster it would have gone alone, and it helped him raise funds to keep his small business going for a little longer, which means my friends are happier and healthier.

@hipsterelectron @fasterandworse my conclusion was that experiment was that it would be faster for me to simply prioritize my recovery and work slower on my projects here and there without the use of "ai" tools at all than it would have been to use them.

@spacehobo I, can, feel, no, sense, of, mea-sure

@hipsterelectron @fasterandworse and when I say it can't do my job, I know this to be true because I evaluated it myself. I have a neck injury, and I have side projects, and I was cautiously hopeful that if these tools worked the way they're marketed to, then I could maybe use them to make working on my side projects safer by reducing the time I spend on them. What I found instead is that these tools are complete garbage.

@hipsterelectron @fasterandworse in the long term, "ai" can't and will not be able to do my job, and no amount of hype or magical thinking on the part of the leisure class will ever change that, because we don't live in a consensus reality and magical thinking can't magick away the limitations of the technology. Meanwhile, there's been a steady stream of research showing that there's a rather large gap between the actual reality of the situation and what the "ai" peddlers are saying.