4 ms·

> Now I just point out behavior that I want to be different and the model figures it out. Often surprisingly quickly. This seems to have low-to-zero value in t

by beej71 4mo ago

> Now I just point out behavior that I want to be different and the model figures it out. Often surprisingly quickly. This seems to have low-to-zero value in the medium- to long-term. Probably some 90% of the population has the ability to sit in their living room and ask a genAI to build things. Why do I need to use your software? If I liked it, I'd just have my AI clone it. And if everyone can just groom dogs (using this to mean "non-information work"), the value of that goes through the floor, too. > I see AIs as providing quality (better) and humans as providing flavor Humans are perhaps better than AI at providing flavor? > I see myself now way more useful as contributing unique ideas of what could exist and how (on the high level) it could work. Another example of being better. > Except perhaps a bit of taste when it comes to directory layout and dependencies selection. And another. > I'm sure this comment would be an easier read if the bot read it for you and relied the ideas and information it contains I'm not sure that's the case, but if it were and that's what I was doing, there'd be no point in human flavor. :(

scotty79 4mo ago
> This seems to have low-to-zero value in the medium- to long-term. On one hand, I don't think so. Because idea guys seemed to be doing well even in a world where it's super expensive to convert the idea into anything of value. On the other regardless of what you do you can't earn money if the customer for whatever you are trying to sell has no money to pay you and that might be the case if a lot of people lose their income stream and nobody replaces it with something else. > Probably some 90% of the population has the ability to sit in their living room and ask a genAI to build things. Why do I need to use your software? If I liked it, I'd just have my AI clone it. Yes. And that's the beauty of it. You won't clone my software. Because it's mine. I won't even show it to you, let alone sell it or give it to you. But I'll happily sit in your living room with you to help you build one for yourself. Provided that you'll compensate me for my time and effort. Most software of the future will have userbase of 1. Everybody can exercise, yet people hire trainers. Everyone can live, yet people hire couches for everything imaginable. Can AI be a couch? Some day, but people stubbornly like human touch in that role. > And if everyone can just groom dogs (using this to mean "non-information work"), the value of that goes through the floor, too. My point is, jobs can be invented out of thin air if there's money around to spend. And they were. And they will be. > Humans are perhaps better than AI at providing flavor? Yes. And it might stay like that for a while longer. Just because we have billions of somewhat unique brains in stock. >> Except perhaps a bit of taste when it comes to directory layout and dependencies selection. > And another. Am I to take pride in my skill of directory naming? Am I to tie my self-worth with that? I don't think that's prudent. It will go away soon anyways. I'm already learning new worthy dependencies from what AI built for me. And it does quite reasonable directory naming when prompted already. >> I'm sure this comment would be an easier read if the bot read it for you and relied the ideas and information it contains > I'm not sure that's the case, You can try. Don't be afraid to experiment. It might be bad, but you might have ideas how to make it a bit better. You can tell agent your ideas and see how they pan out. > but if it were and that's what I was doing, there'd be no point in human flavor. If the human has interesting flavor to provide it tends to show up even in the finished product. Regardless of whether machine that built it was made of people and processes or just weights and loops.
- beej71 4mo ago
  > Because idea guys seemed to be doing well even in a world where it's super expensive to convert the idea into anything of value. But now we're talking about a world where it's super cheap to do that. > But I'll happily sit in your living room with you to help you build one for yourself. Provided that you'll compensate me for my time and effort. I'd pay the market rate, which would be basically zero due to oversupply. > but people stubbornly like human touch in that role. A point of view I can definitely relate to. :) "ChatGPT, please summarize the heartfelt message my aging mother sent me." Students also tell me this. When I discussed the whole idea of AI teacher "avatars" with them, their main feedback was that it was extremely disrespectful to them. They know how to use AI. They came to school to learn more than AI could teach them, from people who are better at teaching than AI is. Admittedly, genAI is a pretty goddamn good teacher. But the best human teachers are better. And man, I would definitely prefer to read human-produced material, but it's really hard to separate the wheat from the chaff today. > My point is, jobs can be invented out of thin air if there's money around to spend. And they were. And they will be. I agree with you, but those jobs don't necessarily have any value if there's an oversupply of workers. And this is especially true for jobs that are easy to learn and perform. Groundskeeping will pay zero. Food prep will pay zero. Even trades that are more challenging to learn (like electrical or welding) will pay zero when 20% of the population is a welder. So I have the same question: where is the money going to come from? No money is economic collapse. > Am I to take pride in my skill of directory naming? Am I to tie my self-worth with that? I don't think that's prudent. Personally, at this point, my recommendation is that you tie your self-worth (when it comes to monetary production ability) to anything and everything you possibly can. Tie it to anything that gives you any edge whatsoever. > Don't be afraid to experiment. It might be bad, but you might have ideas how to make it a bit better. You can tell agent your ideas and see how they pan out. Yeah, I've done a lot of this. I'm impressed at how well it can get nearly up to par, but really unimpressed by how poorly it exceeds it. I mean, it's good. But it's not standout. But in this case, I don't like how it would strip away the you part of your message. If all I cared about was raw information, I wouldn't even be here now. I'd just have my secretary-bot summarize and write a response. I want this conversation to be with you, human scotty79. What is your job, out of curiosity, if you're willing to share? And if it's tech, what do you plan to do after that market is gone?

by beej71 4mo ago

scotty79 4mo ago

> This seems to have low-to-zero value in the medium- to long-term. On one hand, I don't think so. Because idea guys seemed to be doing well even in a world where it's super expensive to convert the idea into anything of value. On the other regardless of what you do you can't earn money if the customer for whatever you are trying to sell has no money to pay you and that might be the case if a lot of people lose their income stream and nobody replaces it with something else. > Probably some 90% of the population has the ability to sit in their living room and ask a genAI to build things. Why do I need to use your software? If I liked it, I'd just have my AI clone it. Yes. And that's the beauty of it. You won't clone my software. Because it's mine. I won't even show it to you, let alone sell it or give it to you. But I'll happily sit in your living room with you to help you build one for yourself. Provided that you'll compensate me for my time and effort. Most software of the future will have userbase of 1. Everybody can exercise, yet people hire trainers. Everyone can live, yet people hire couches for everything imaginable. Can AI be a couch? Some day, but people stubbornly like human touch in that role. > And if everyone can just groom dogs (using this to mean "non-information work"), the value of that goes through the floor, too. My point is, jobs can be invented out of thin air if there's money around to spend. And they were. And they will be. > Humans are perhaps better than AI at providing flavor? Yes. And it might stay like that for a while longer. Just because we have billions of somewhat unique brains in stock. >> Except perhaps a bit of taste when it comes to directory layout and dependencies selection. > And another. Am I to take pride in my skill of directory naming? Am I to tie my self-worth with that? I don't think that's prudent. It will go away soon anyways. I'm already learning new worthy dependencies from what AI built for me. And it does quite reasonable directory naming when prompted already. >> I'm sure this comment would be an easier read if the bot read it for you and relied the ideas and information it contains > I'm not sure that's the case, You can try. Don't be afraid to experiment. It might be bad, but you might have ideas how to make it a bit better. You can tell agent your ideas and see how they pan out. > but if it were and that's what I was doing, there'd be no point in human flavor. If the human has interesting flavor to provide it tends to show up even in the finished product. Regardless of whether machine that built it was made of people and processes or just weights and loops.

beej71 4mo ago
> Because idea guys seemed to be doing well even in a world where it's super expensive to convert the idea into anything of value. But now we're talking about a world where it's super cheap to do that. > But I'll happily sit in your living room with you to help you build one for yourself. Provided that you'll compensate me for my time and effort. I'd pay the market rate, which would be basically zero due to oversupply. > but people stubbornly like human touch in that role. A point of view I can definitely relate to. :) "ChatGPT, please summarize the heartfelt message my aging mother sent me." Students also tell me this. When I discussed the whole idea of AI teacher "avatars" with them, their main feedback was that it was extremely disrespectful to them. They know how to use AI. They came to school to learn more than AI could teach them, from people who are better at teaching than AI is. Admittedly, genAI is a pretty goddamn good teacher. But the best human teachers are better. And man, I would definitely prefer to read human-produced material, but it's really hard to separate the wheat from the chaff today. > My point is, jobs can be invented out of thin air if there's money around to spend. And they were. And they will be. I agree with you, but those jobs don't necessarily have any value if there's an oversupply of workers. And this is especially true for jobs that are easy to learn and perform. Groundskeeping will pay zero. Food prep will pay zero. Even trades that are more challenging to learn (like electrical or welding) will pay zero when 20% of the population is a welder. So I have the same question: where is the money going to come from? No money is economic collapse. > Am I to take pride in my skill of directory naming? Am I to tie my self-worth with that? I don't think that's prudent. Personally, at this point, my recommendation is that you tie your self-worth (when it comes to monetary production ability) to anything and everything you possibly can. Tie it to anything that gives you any edge whatsoever. > Don't be afraid to experiment. It might be bad, but you might have ideas how to make it a bit better. You can tell agent your ideas and see how they pan out. Yeah, I've done a lot of this. I'm impressed at how well it can get nearly up to par, but really unimpressed by how poorly it exceeds it. I mean, it's good. But it's not standout. But in this case, I don't like how it would strip away the you part of your message. If all I cared about was raw information, I wouldn't even be here now. I'd just have my secretary-bot summarize and write a response. I want this conversation to be with you, human scotty79. What is your job, out of curiosity, if you're willing to share? And if it's tech, what do you plan to do after that market is gone?