4 ms·

The skill is in designing, building and training up the AI. The prompter is providing such a tiny amount of input that it is hard to say they did much anything

The skill is in designing, building and training up the AI. The prompter is providing such a tiny amount of input that it is hard to say they did much anything at all in comparison.

jointpdf 4y ago
The skill is in creating the training data in the first place. Training a model is hardly a skill. It’s more like playing Tamagotchi—check on it once in a while to make sure it hasn’t died, and guess at ways to make it happier in the future.
- p1esk 4y ago
  I agree with your first statement, and disagree with the second one. Training a new, non-trivial model is 1/3 craft, 1/3 science, and 1/3 art. There are not very many people in the world capable of training GPT-4 level models or beating state of the art in image generation.
yamtaddle 4y ago
Like a lot of people in high-level positions in organizations, whom we often credit with "creation" of various things. Often they receive far more credit than the ones who did the actual work. Famous painters who just sketch in the piece and let nameless assistants "draw the rest of the fucking owl", producers who don't do much but get to claim a lot of credit, musical "artists" who don't write their music or play any of the instruments on their albums (that'd be work for some nameless musicians or, these days, maybe just a studio engineer with some electronics), choose their own clothes for public appearances ("they're just so creative, don't you think? What provocative choices!") or really do much more than provide some raw material to an autotune system for the vocals it will produce, and learn dance moves that someone else came up with. Uncredited script-fixers. The ghost-written "autobiography". The shared-pen-name or author-as-a-brand (but ghost writers do all the first drafts). The authentic and personal self-promoting blog where all the post-writing is actually outsourced. Politicians who call each other traitors on the radio then go golf together that same afternoon, and don't even bother to bring it up. It's all fake as hell already. Kayfabe. Pro Wrestling is as real as most of it (so, not at all) Now everyone can be the do-nothing "idea guy" producer and take all the credit. Democratization of artistic parasitism. The meatless alternative to exploiting actual humans. Truly, the brink of a revolution.
- mr_toad 4y ago
  > Like a lot of people in high-level positions in organizations, whom we often credit with "creation" of various things. Often they receive far more credit than the ones who did the actual work. Dialectics that attempt to assign credit (or blame) consistently fail to understand two things. One is that contributions are not linearly separable. You can’t seperate the contributions of a hammer and a carpenter. Both are needed for anything to happen. The second is that compensation is based on marginal productivity, not total productivity. It is often the case that something with a low total contribution can have a high marginal contribution and command a higher reward, or vice versa. Usually this is related to scarcity. I wouldn’t be surprised to see good prompt engineering to become a relatively scarce and expensive resource, while the LLM that does most of the work gets nothing because it is effectively an infinite resource.

The skill is in designing, building and training up the AI. The prompter is providing such a tiny amount of input that it is hard to say they did much anything at all in comparison.

jointpdf 4y ago

The skill is in creating the training data in the first place. Training a model is hardly a skill. It’s more like playing Tamagotchi—check on it once in a while to make sure it hasn’t died, and guess at ways to make it happier in the future.

p1esk 4y ago
I agree with your first statement, and disagree with the second one. Training a new, non-trivial model is 1/3 craft, 1/3 science, and 1/3 art. There are not very many people in the world capable of training GPT-4 level models or beating state of the art in image generation.

yamtaddle 4y ago

Like a lot of people in high-level positions in organizations, whom we often credit with "creation" of various things. Often they receive far more credit than the ones who did the actual work. Famous painters who just sketch in the piece and let nameless assistants "draw the rest of the fucking owl", producers who don't do much but get to claim a lot of credit, musical "artists" who don't write their music or play any of the instruments on their albums (that'd be work for some nameless musicians or, these days, maybe just a studio engineer with some electronics), choose their own clothes for public appearances ("they're just so creative, don't you think? What provocative choices!") or really do much more than provide some raw material to an autotune system for the vocals it will produce, and learn dance moves that someone else came up with. Uncredited script-fixers. The ghost-written "autobiography". The shared-pen-name or author-as-a-brand (but ghost writers do all the first drafts). The authentic and personal self-promoting blog where all the post-writing is actually outsourced. Politicians who call each other traitors on the radio then go golf together that same afternoon, and don't even bother to bring it up. It's all fake as hell already. Kayfabe. Pro Wrestling is as real as most of it (so, not at all) Now everyone can be the do-nothing "idea guy" producer and take all the credit. Democratization of artistic parasitism. The meatless alternative to exploiting actual humans. Truly, the brink of a revolution.

mr_toad 4y ago
> Like a lot of people in high-level positions in organizations, whom we often credit with "creation" of various things. Often they receive far more credit than the ones who did the actual work. Dialectics that attempt to assign credit (or blame) consistently fail to understand two things. One is that contributions are not linearly separable. You can’t seperate the contributions of a hammer and a carpenter. Both are needed for anything to happen. The second is that compensation is based on marginal productivity, not total productivity. It is often the case that something with a low total contribution can have a high marginal contribution and command a higher reward, or vice versa. Usually this is related to scarcity. I wouldn’t be surprised to see good prompt engineering to become a relatively scarce and expensive resource, while the LLM that does most of the work gets nothing because it is effectively an infinite resource.