Generative AI is a Faustian bargain.
-
@benlockwood electrically powered tools are a faustian bargain. Every use of them weakens you...
Electric drill diminishes your use of muscles that could've been used for hammering nails manually
@tylnesh @benlockwood
Oof. I appreciate your intention about AI and agree with that intention.
But this was a poor choice for the simile. I work with crew almost every day. In most cases, power tools are a way of avoiding repetitive stress injuries and have nothing to do with sparing muscle fatigue. They are a safety measure.
I apologize for the interruption. -
@corigan01 @tylnesh @benlockwood
Also, all these other examples are deterministic systems which can be understood and utilised with agency. LLM output is fundamentally randomised, using it is just rolling some dice and hoping that the result will be useful; you actually do hand off all agency over your own life when doing so.@painting_squirrel @corigan01 @benlockwood I don’t hope, I read, reread, reason about it and verify the output. Also saying that LLMs are non deterministic is a gross simplification. Set the model temperature to 0, greedy decoding and fixed seed and barring floating point inaccuracies and similar influences you get close to deterministic output. If you want to go as close as possible, run inference on a cpu.
Do you use compilers or do you write assembly?
-
Generative AI is a Faustian bargain. Every use of it weakens you. Maybe you occasionally save a few minutes, but you are diminished for it. With each use you are further from your human potential.
I always say that AI is like the One Ring. You think it's serving you, but it really only serves its master.
-
@tylnesh @benlockwood
Oof. I appreciate your intention about AI and agree with that intention.
But this was a poor choice for the simile. I work with crew almost every day. In most cases, power tools are a way of avoiding repetitive stress injuries and have nothing to do with sparing muscle fatigue. They are a safety measure.
I apologize for the interruption.@ohmu @benlockwood power tools often cause RSI, albeit different from manual equivalents.
-
@tylnesh @corigan01 @benlockwood Confusing memorization with actual learning is part of what has gotten us in the mess we are in now.
@rasterweb @corigan01 @benlockwood if you want to be nitpicky, let me change my metaphor to “Learn to navigate by stars, instead of using a compass” yea, it’s very cool and useful (at night), but refusing new ways of doing things because you lack understanding and there are assholes abusing it is giving away power to those assholes.
-
@rasterweb @corigan01 @benlockwood if you want to be nitpicky, let me change my metaphor to “Learn to navigate by stars, instead of using a compass” yea, it’s very cool and useful (at night), but refusing new ways of doing things because you lack understanding and there are assholes abusing it is giving away power to those assholes.
@tylnesh @corigan01 @benlockwood Can you use your own tools and not use the tools the assholes provide and control?
If you are paying the assholes to use their services you need to determine where you fit into the asshole equation.
-
@undead @benlockwood yeah, my solar powered second hand gpu is burning forests down.
-
@benlockwood this comment section has given me so many valuable lessons in blocking before arguing with unaware people. Thank you.
@MamaLake@beige.party @benlockwood@ecoevo.social blocking anti-AI or pro-AI people? Or regardless of their stance, blocking people that used bad arguments and/or know little of what they were defending?
I have an idea but, I'm genuinely curious about your answer. -
Generative AI is a Faustian bargain. Every use of it weakens you. Maybe you occasionally save a few minutes, but you are diminished for it. With each use you are further from your human potential.
@benlockwood @lisamelton This would make a great modern-day fairy tale.
-
@corigan01 @benlockwood If I were resorting to it constantly, I might agree, but I'm not. It's a last resort for really bad days. (And to be clear, I'm not a classic ADHD case; I'm borderline but with other shit going on, as well).
@jack @corigan01 @benlockwood There are people who suffer from depression who talk to chatbots when they have a bad day. They say it helps, and that interacting with a relentlessly friendly bot is more helpful than the difficult task of going out and interacting with humans. I think what they’re doing is incredibly dangerous. These tools lack robust behavior safeguards, and are literally designed to suck you in and make you dependent and addicted. But hey, I hope it works out for you and doesn’t harm you.
-
@benlockwood I don't think that's necessarily true.
Sure if you don't ride the bike every day you're worse bike rider but finishing races also makes you a better bike rider.
I'm a senior dev and consider myself a good'ish coder and while LLMs do make me worse at memorizing APIs I ship so much more which on it's own a very useful skill to train.
At the end of the day it's just another computer. If you use the one that's in your skull as well then you'll be just perfectly fine.
@wraptile @benlockwood ”senior dev and consider myself a good'ish coder” should make you realise that ”I ship so much more” is not a good thing.
-
@benlockwood I don't think that's necessarily true.
Sure if you don't ride the bike every day you're worse bike rider but finishing races also makes you a better bike rider.
I'm a senior dev and consider myself a good'ish coder and while LLMs do make me worse at memorizing APIs I ship so much more which on it's own a very useful skill to train.
At the end of the day it's just another computer. If you use the one that's in your skull as well then you'll be just perfectly fine.
aure parrot
-
Generative AI is a Faustian bargain. Every use of it weakens you. Maybe you occasionally save a few minutes, but you are diminished for it. With each use you are further from your human potential.
Well big question is what are people who used it supposed to do about this (besides just quit)? Those with mental disorders are VERY easily swayed by it, especially those with OCD, who are VERY drawn towards stupid question answering reassurance machine, to the point it becomes its own addiction (ask me how i know... i still cant fully stop myself...). Do you believe this damage can ever be fixed? How do you "get off" this cycle?
-
Well big question is what are people who used it supposed to do about this (besides just quit)? Those with mental disorders are VERY easily swayed by it, especially those with OCD, who are VERY drawn towards stupid question answering reassurance machine, to the point it becomes its own addiction (ask me how i know... i still cant fully stop myself...). Do you believe this damage can ever be fixed? How do you "get off" this cycle?
@poor_labsafety I truly cannot relate and have no idea what to tell you. I have no desire whatsoever to ask a machine random questions and can’t fathom doing so.
-
I always say that AI is like the One Ring. You think it's serving you, but it really only serves its master.
@MegaMichelle @benlockwood
"AI" is doing a lot of heavy lifting in that sentence. A decision tree classifying tumors, a reinforcement learning agent balancing a robot arm, a recommendation system none of those have a "master." They're narrow tools doing one job. The "serving its master" worry only applies to one slice: generative models trained on someone else's data by a company you don't work for. Using "AI" to mean "that one scary thing" is like saying vehicles are evil because of that one truck that ran over me.. think about it and know the distinction between the whole of AI and Corporation led generative AI please
"AI is the One Ring" Cool metaphor until you realize the One Ring is one object, one master, and it's magic you can't inspect. An LLM is a model file on your disk. You can read its weights, run it offline, point it at your own data, and delete it. The Ring doesn't let you do that. The Ring also talks to Sauron. Your local LLM doesn't talk to anyone. You built it. -
I always say that AI is like the One Ring. You think it's serving you, but it really only serves its master.
@MegaMichelle @benlockwood sharing just in case you haven't seen this yet
-
@MegaMichelle @benlockwood
"AI" is doing a lot of heavy lifting in that sentence. A decision tree classifying tumors, a reinforcement learning agent balancing a robot arm, a recommendation system none of those have a "master." They're narrow tools doing one job. The "serving its master" worry only applies to one slice: generative models trained on someone else's data by a company you don't work for. Using "AI" to mean "that one scary thing" is like saying vehicles are evil because of that one truck that ran over me.. think about it and know the distinction between the whole of AI and Corporation led generative AI please
"AI is the One Ring" Cool metaphor until you realize the One Ring is one object, one master, and it's magic you can't inspect. An LLM is a model file on your disk. You can read its weights, run it offline, point it at your own data, and delete it. The Ring doesn't let you do that. The Ring also talks to Sauron. Your local LLM doesn't talk to anyone. You built it.@zer0unplanned @benlockwood @MegaMichelle no, the useful things already had a name for them, the name is "machine learning". The name for the various surveillance slop machines is AI. Don't mix the two.
-
Well big question is what are people who used it supposed to do about this (besides just quit)? Those with mental disorders are VERY easily swayed by it, especially those with OCD, who are VERY drawn towards stupid question answering reassurance machine, to the point it becomes its own addiction (ask me how i know... i still cant fully stop myself...). Do you believe this damage can ever be fixed? How do you "get off" this cycle?
@poor_labsafety
I suggest remembering that this stuff has only existed for a very short time and you got along perfectly fine without it all your life before you touched it. So you can again. And that there are millions and millions of people who have no interest in it and get along perfectly fine (actually even better!) without it. [contemplating this is one of the ways that helped me quit smoking].Distraction is another technique. Do something else that engages your mind. For example if you have a video game you like to play, whenever you feel the need to talk to the bot, play the game for a while instead.
The mind is like a happy child who sees a toy at the store and is told no, he can't have it. He instantly becomes miserable because he feels such a strong need to have the toy. Yet before entering the store he was perfectly happy, and will be again once enough time has passed without the toy.
-
@MegaMichelle @benlockwood
"AI" is doing a lot of heavy lifting in that sentence. A decision tree classifying tumors, a reinforcement learning agent balancing a robot arm, a recommendation system none of those have a "master." They're narrow tools doing one job. The "serving its master" worry only applies to one slice: generative models trained on someone else's data by a company you don't work for. Using "AI" to mean "that one scary thing" is like saying vehicles are evil because of that one truck that ran over me.. think about it and know the distinction between the whole of AI and Corporation led generative AI please
"AI is the One Ring" Cool metaphor until you realize the One Ring is one object, one master, and it's magic you can't inspect. An LLM is a model file on your disk. You can read its weights, run it offline, point it at your own data, and delete it. The Ring doesn't let you do that. The Ring also talks to Sauron. Your local LLM doesn't talk to anyone. You built it.@zer0unplanned
Like it or not, 'AI' now means LLMs/generative AI because the broligarchs adopted the term and made it so. It's a fait accompli.If it once meant something different to some people, that's over now. Any efforts to change that by replying to posts on Mastodon are as futile as trying to bail out the ocean with a bucket.
-
@wraptile @benlockwood ”senior dev and consider myself a good'ish coder” should make you realise that ”I ship so much more” is not a good thing.
@ahltorp why is it not a good thing?
