How did worry about AI safety become a tech bro signifier and not the opposite? It’s clear that the world as we know it only has a few years left but people keep their heads in the sand because they can’t really grasp the idea of exponential progress. Your mental model of ai is probably something like the 2024 freemium ChatGPT models that couldn’t count the r’s in strawberry. But these things get better by the month and remember the anthropological rule that it’s impossible to un-invent something. If you express this in cultured company though they’ll look at you like you’re some sort of thielite San franciscan. Why?
Anonymous :
55 days ago :
No.10697
>>10698
>>10697
the important part isn’t really the improvements themselves it’s the improvement in the speed of improvement. 2nd derivative like Reagan talked about. eg: it was like 2 years for AI images to figure out fingers, but we got from there to basically photorealistic video in far less time. But if you’re interested in the actual capabilities the “system cards” Anthropic publishes on their site are informative. They have benchmarks on bioweapons, cyber warfare, subjective experience, etc, that change with each new release
What are the key advances in the last year? Claude Code has gotten better, right? And it solved a maths theorem...?
>>10697
What are the key advances in the last year? Claude Code has gotten better, right? And it solved a maths theorem...?
the important part isn’t really the improvements themselves it’s the improvement in the speed of improvement. 2nd derivative like Reagan talked about. eg: it was like 2 years for AI images to figure out fingers, but we got from there to basically photorealistic video in far less time. But if you’re interested in the actual capabilities the “system cards” Anthropic publishes on their site are informative. They have benchmarks on bioweapons, cyber warfare, subjective experience, etc, that change with each new release
Anonymous :
55 days ago :
No.10699
>>10700
>>10699
Very eloquent comment but I think you have a serious lack of imagination if those are the only consequences you can envision. and of course it’d activate your “crazy radar”; the existence of human level artificial intelligence should already have done that!
Like many things, AI safety got ruined by the stench of spergery; in this case all that Berkeley basilisk bullshit. People haven't "updated their priors" because life hasn't changed that much for the chattering class. Ultimately we keep Graeberjobs or we don't. I don't see much evidence for them going away entirely. (Yes the pie will continue to shrink.)
>I'm very specific when expressing my grievances and everyone listening understands that I'm talking about [the future of art and culture]
Not that many people care about art or culture.
>[the political consequences of on-demand zone flooding]
People can't really get much more entrenched (this is actually the exponentiation I'm most worried about right now) thanks to continued Blueanon acquiescence/whatever the fuck brown Hitlerites are getting on short-form. I know AI has contributed to this but I really think it's a matter of a full tank vs. 1 or 2 used gallons.
IMO the more things change the more they stay the same. I could have my head in the sand but "the world as we know it only has a few years left" is never gonna get great responses because it activates people's crazy radars
Anonymous :
55 days ago :
No.10700
>>10705
>>10700
I'm open to your imagination - I agree mine is limited - but you have to be more specific. People are responding to you as a tech bro because framings like "the existence of human level artifical intelligence" are mostly made by tech bros. Chess fell 30 years ago, Madden 10. I wouldn't have a degree without Mathematica. Rasterization followed the same curve you described, what was the last game you bought? Was it fun?
>>10699
Like many things, AI safety got ruined by the stench of spergery; in this case all that Berkeley basilisk bullshit. People haven't "updated their priors" because life hasn't changed that much for the chattering class. Ultimately we keep Graeberjobs or we don't. I don't see much evidence for them going away entirely. (Yes the pie will continue to shrink.)
>I'm very specific when expressing my grievances and everyone listening understands that I'm talking about [the future of art and culture]
Not that many people care about art or culture.
>[the political consequences of on-demand zone flooding]
People can't really get much more entrenched (this is actually the exponentiation I'm most worried about right now) thanks to continued Blueanon acquiescence/whatever the fuck brown Hitlerites are getting on short-form. I know AI has contributed to this but I really think it's a matter of a full tank vs. 1 or 2 used gallons.
IMO the more things change the more they stay the same. I could have my head in the sand but "the world as we know it only has a few years left" is never gonna get great responses because it activates people's crazy radars
Very eloquent comment but I think you have a serious lack of imagination if those are the only consequences you can envision. and of course it’d activate your “crazy radar”; the existence of human level artificial intelligence should already have done that!
Anonymous :
55 days ago :
No.10705
>>10713
>>10705
If any job which doesn’t involve physical interface with the world can be run off a bank of servers in Utah, it doesn’t matter whether the tokens cost thousands of dollars a month as long as it’s cheaper than paying a human being 6 figures a year. No good capitalist is going to hold on to human employees out of sentimentality when there’s that much extra surplus value to vacuum up. So that should have some economic ramifications.
Another one is the fact that AI allows for basically perfect surveillance and maybe even minority-report level prevention if it was actually deployed fully. A couple more Roblox-grown school shooters would probobly give legislators a good enough excuse to do that. There’s also the fact that the internet would become completely unusable (although it’s debatable whether the internet of discreet websites would still be a thing with good enough LLMs) because it would be literally impossible to determine whether someone was a human or robot without some form of ID verification, which would fundamentally change the internet as well. Those things and their downstream effects seem pretty radical, and that’s not even getting into war/government/economy stuff
>>10700
>>10699
Very eloquent comment but I think you have a serious lack of imagination if those are the only consequences you can envision. and of course it’d activate your “crazy radar”; the existence of human level artificial intelligence should already have done that!
I'm open to your imagination - I agree mine is limited - but you have to be more specific. People are responding to you as a tech bro because framings like "the existence of human level artifical intelligence" are mostly made by tech bros. Chess fell 30 years ago, Madden 10. I wouldn't have a degree without Mathematica. Rasterization followed the same curve you described, what was the last game you bought? Was it fun?
I think you’re just expecting way too much from humanities people. They think of things in terms of broad comparisons (to crypto hype, to 2001 and 2007 crashes) and have no interest in the lesswrong thought experiment stuff. There’s also a nonzero chance that they’re right, but I wouldn’t bet on it
Anonymous :
55 days ago :
No.10707
>>10708
>>10707
The image generation AIs do not matter for any of this. They’re architecturally and strategically unrelated to LLMs
>>10712>>10707
I'm getting straight prose from chatGPT whenever I ask it something, and I haven't given it anything to remember. Its responses are generally sensible, though I know what types of questions result in useful output, so I don't use it outside of those cases.
thing is that most people using ai are seeing a downgrade in llm's responses. gpt when it comes to quality of responses has definitely gotten worse. it peaked when it went mainstream and now it gives you emoji slop bullet points slop unless you put a lot of effort into customising it to not do that. it hallucinated when it went mainstream and it still does now. only thing that's gotten better than, to the average person is how much more realistic ai pictures look. meanwhile behind the scenes all these ai companies have already improved the models in ways I can't even imagine and are granting access to it at incomprehensible costs to ultra rich people. that's my two cents on what I think is happening
>>10707
thing is that most people using ai are seeing a downgrade in llm's responses. gpt when it comes to quality of responses has definitely gotten worse. it peaked when it went mainstream and now it gives you emoji slop bullet points slop unless you put a lot of effort into customising it to not do that. it hallucinated when it went mainstream and it still does now. only thing that's gotten better than, to the average person is how much more realistic ai pictures look. meanwhile behind the scenes all these ai companies have already improved the models in ways I can't even imagine and are granting access to it at incomprehensible costs to ultra rich people. that's my two cents on what I think is happening
The image generation AIs do not matter for any of this. They’re architecturally and strategically unrelated to LLMs
>>10707
thing is that most people using ai are seeing a downgrade in llm's responses. gpt when it comes to quality of responses has definitely gotten worse. it peaked when it went mainstream and now it gives you emoji slop bullet points slop unless you put a lot of effort into customising it to not do that. it hallucinated when it went mainstream and it still does now. only thing that's gotten better than, to the average person is how much more realistic ai pictures look. meanwhile behind the scenes all these ai companies have already improved the models in ways I can't even imagine and are granting access to it at incomprehensible costs to ultra rich people. that's my two cents on what I think is happening
I'm getting straight prose from chatGPT whenever I ask it something, and I haven't given it anything to remember. Its responses are generally sensible, though I know what types of questions result in useful output, so I don't use it outside of those cases.
>>10705
>>10700
I'm open to your imagination - I agree mine is limited - but you have to be more specific. People are responding to you as a tech bro because framings like "the existence of human level artifical intelligence" are mostly made by tech bros. Chess fell 30 years ago, Madden 10. I wouldn't have a degree without Mathematica. Rasterization followed the same curve you described, what was the last game you bought? Was it fun?
If any job which doesn’t involve physical interface with the world can be run off a bank of servers in Utah, it doesn’t matter whether the tokens cost thousands of dollars a month as long as it’s cheaper than paying a human being 6 figures a year. No good capitalist is going to hold on to human employees out of sentimentality when there’s that much extra surplus value to vacuum up. So that should have some economic ramifications.
Another one is the fact that AI allows for basically perfect surveillance and maybe even minority-report level prevention if it was actually deployed fully. A couple more Roblox-grown school shooters would probobly give legislators a good enough excuse to do that. There’s also the fact that the internet would become completely unusable (although it’s debatable whether the internet of discreet websites would still be a thing with good enough LLMs) because it would be literally impossible to determine whether someone was a human or robot without some form of ID verification, which would fundamentally change the internet as well. Those things and their downstream effects seem pretty radical, and that’s not even getting into war/government/economy stuff
Anonymous :
28 days ago :
No.10981
>>10996
>>10981
you can’t even call it a frog boiling thing because it’s not happening slowly at all. It’s more like a frog not noticing it’s being blowtorched
the inability to consider the ills, or the wholesale adoption, of AI by the average Joe has successfully buck broken me, and I will just lay low to the ground from here on out
>>10981
the inability to consider the ills, or the wholesale adoption, of AI by the average Joe has successfully buck broken me, and I will just lay low to the ground from here on out
you can’t even call it a frog boiling thing because it’s not happening slowly at all. It’s more like a frog not noticing it’s being blowtorched
Anonymous :
25 days ago :
No.11056
>>11061
>>11056
>AI seems like a really cool area of research that should've stayed within the confines of university campuses and research labs until it was ready to be actually used somewhere.
Doesn't sound too dissimilar to the internet, or other "destructive" technologies.
>>11068>>11056
>What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
yes? what's confusing about that
>exponential progress
There is literally no evidence that AI improvement is exponential. The marketing teams of these companies seem to be borrowing that idea from Moore's Law which applies to hardware (not software) and has been dead since the 2010. Jensen Huang himself stated that he considers it dead. We've also had tons of AI/Automaton scares throughout history, this is not a new concept or grift. Additionally, CNNs have existed for a long time and AlexNet came out almost 15 years ago.
Truthfully, I know next to nothing about AI even though I'm a software engineer, the classes I took in it were before ChatGPT dropped. I truly hate the technology and think it's a step backwards for most of the industry. The actual use cases for it are extremely narrow, the vast majority of people and businesses use computers because data is reliable and 100% reproducible. I can't think of many situations where you need unreliable results that are generated in a proprietary blackbox. Human employees have a lot more value than just the tasks they complete, specifically they can be held accountable for mistakes. What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
I just don't see a world where general AI models are doing anything. Even something as trivial as Flock Cameras are getting insane pushback and violate the constitution. AI seems like a really cool area of research that should've stayed within the confines of university campuses and research labs until it was ready to be actually used somewhere.
>How did worry about AI safety become a tech bro signifier and not the opposite?
Most liberal arts majors aren't in a tech echo-chamber and they don't stand to get 7+ figure payouts from AI taking off. Worrying about "AI safety" is just so stupid the only people who talk about that are either autistic nerds who could be manipulated into believing anything or people who stand to make a profit.
If you think ChatGPT "broke containment" you're probably sitting with sub 90IQ and I'm not even joking.
>>11056
>exponential progress
There is literally no evidence that AI improvement is exponential. The marketing teams of these companies seem to be borrowing that idea from Moore's Law which applies to hardware (not software) and has been dead since the 2010. Jensen Huang himself stated that he considers it dead. We've also had tons of AI/Automaton scares throughout history, this is not a new concept or grift. Additionally, CNNs have existed for a long time and AlexNet came out almost 15 years ago.
Truthfully, I know next to nothing about AI even though I'm a software engineer, the classes I took in it were before ChatGPT dropped. I truly hate the technology and think it's a step backwards for most of the industry. The actual use cases for it are extremely narrow, the vast majority of people and businesses use computers because data is reliable and 100% reproducible. I can't think of many situations where you need unreliable results that are generated in a proprietary blackbox. Human employees have a lot more value than just the tasks they complete, specifically they can be held accountable for mistakes. What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
I just don't see a world where general AI models are doing anything. Even something as trivial as Flock Cameras are getting insane pushback and violate the constitution. AI seems like a really cool area of research that should've stayed within the confines of university campuses and research labs until it was ready to be actually used somewhere.
>How did worry about AI safety become a tech bro signifier and not the opposite?
Most liberal arts majors aren't in a tech echo-chamber and they don't stand to get 7+ figure payouts from AI taking off. Worrying about "AI safety" is just so stupid the only people who talk about that are either autistic nerds who could be manipulated into believing anything or people who stand to make a profit.
If you think ChatGPT "broke containment" you're probably sitting with sub 90IQ and I'm not even joking.
>AI seems like a really cool area of research that should've stayed within the confines of university campuses and research labs until it was ready to be actually used somewhere.
Doesn't sound too dissimilar to the internet, or other "destructive" technologies.
Anonymous :
25 days ago :
No.11068
>>11069
>>11068
What I'm getting that is that either situation makes gen AI impossible to use in business. Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent, or you can't sue them in which case most businesses will opt to not use their products or to at least limit their usage.
>>No.11061
Yeah that's my point, the slow gradual adoption of the internet meant that it was filling actual demand for it. People wanted internet access and willingly paid for it and organically supported its growth. Keeping it in universities and research labs at the beginning helped refine the tech before it was available for consumers.
AI is getting pushed to market way too early, it's like if back in the 40s they decided to build massive 60 acre ENIACs for everyone to take turns with.
>>11056
>exponential progress
There is literally no evidence that AI improvement is exponential. The marketing teams of these companies seem to be borrowing that idea from Moore's Law which applies to hardware (not software) and has been dead since the 2010. Jensen Huang himself stated that he considers it dead. We've also had tons of AI/Automaton scares throughout history, this is not a new concept or grift. Additionally, CNNs have existed for a long time and AlexNet came out almost 15 years ago.
Truthfully, I know next to nothing about AI even though I'm a software engineer, the classes I took in it were before ChatGPT dropped. I truly hate the technology and think it's a step backwards for most of the industry. The actual use cases for it are extremely narrow, the vast majority of people and businesses use computers because data is reliable and 100% reproducible. I can't think of many situations where you need unreliable results that are generated in a proprietary blackbox. Human employees have a lot more value than just the tasks they complete, specifically they can be held accountable for mistakes. What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
I just don't see a world where general AI models are doing anything. Even something as trivial as Flock Cameras are getting insane pushback and violate the constitution. AI seems like a really cool area of research that should've stayed within the confines of university campuses and research labs until it was ready to be actually used somewhere.
>How did worry about AI safety become a tech bro signifier and not the opposite?
Most liberal arts majors aren't in a tech echo-chamber and they don't stand to get 7+ figure payouts from AI taking off. Worrying about "AI safety" is just so stupid the only people who talk about that are either autistic nerds who could be manipulated into believing anything or people who stand to make a profit.
If you think ChatGPT "broke containment" you're probably sitting with sub 90IQ and I'm not even joking.
>What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
yes? what's confusing about that
Anonymous :
25 days ago :
No.11069
>>11070
>>11069
I'm not so sure about other sectors but pretty much everyone in tech uses AI on a more-or-less daily basis. I don't love it, I don't look forward to the day when I don't get to write my own code anymore because I'm obliged to get a robot to do it, but the claim that it's too immature to be useful just doesn't hold water.
At the same time as far as I know it has basically obliterated most of the lower-end of the market for things like translation and copy-editing. Those ENIACs had nothing like that sort of impact on the economy.
>>11074>>11069
1) I meant the Internet should still be kept to the academic sphere
2) Why do you think AI is being pushed to market so aggressively?
>>11083>>11069
>Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent,
nigga what? this is how all companies work!! avoiding lawsuits is an incentive to make sure your product works. why is it any different for AI than a company that makes cars or software or baby cribs? you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
>>11068
>>11056
>What are you gonna do if you AI sales rep gets tricked into selling your product for $1 instead of $50 per unit? Are you gonna sue Anthropic?
yes? what's confusing about that
What I'm getting that is that either situation makes gen AI impossible to use in business. Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent, or you can't sue them in which case most businesses will opt to not use their products or to at least limit their usage.
>>No.11061
Yeah that's my point, the slow gradual adoption of the internet meant that it was filling actual demand for it. People wanted internet access and willingly paid for it and organically supported its growth. Keeping it in universities and research labs at the beginning helped refine the tech before it was available for consumers.
AI is getting pushed to market way too early, it's like if back in the 40s they decided to build massive 60 acre ENIACs for everyone to take turns with.
>>11069
>>11068
What I'm getting that is that either situation makes gen AI impossible to use in business. Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent, or you can't sue them in which case most businesses will opt to not use their products or to at least limit their usage.
>>No.11061
Yeah that's my point, the slow gradual adoption of the internet meant that it was filling actual demand for it. People wanted internet access and willingly paid for it and organically supported its growth. Keeping it in universities and research labs at the beginning helped refine the tech before it was available for consumers.
AI is getting pushed to market way too early, it's like if back in the 40s they decided to build massive 60 acre ENIACs for everyone to take turns with.
I'm not so sure about other sectors but pretty much everyone in tech uses AI on a more-or-less daily basis. I don't love it, I don't look forward to the day when I don't get to write my own code anymore because I'm obliged to get a robot to do it, but the claim that it's too immature to be useful just doesn't hold water.
At the same time as far as I know it has basically obliterated most of the lower-end of the market for things like translation and copy-editing. Those ENIACs had nothing like that sort of impact on the economy.
Anonymous :
25 days ago :
No.11074
>>11086
>>11074
>1)
Sorry I misunderstood
>2)
Probably a hot take think this is almost exclusively Sam Altman's fault. OpenAI made sense remaining a nonprofit with access to literal trillions from investors like Musk and AWS. I think Elon (as divisive as he is) understood the business model just wasn't there yet.
I believe Altman is a sociopath that mimics people whose power/influence/wealth he desires, similar to Elizabeth Holmes. I forget the details and they've been mostly scrubbed from the internet, but he essentially slept his way into his position at YC and leveraged that to get his job as CEO of OpenAI. His whole come up is really strange and built on bullshit, Loopt had like 0 users when it was bought.
He wanted to be a billionaire and rushed an impressive but premature technology to market because he knew it was good enough to scare people into investing. This caused a domino effect where basically every tech company had to buy in or risk scaring their shareholders or missing out on a revolutionary technology (nobody wants to repeat what Ballmer did with the iPhone). However, people aren't considering the alternative situation where maybe this technology isn't like the iPhone.
I want to be clear that I don't think AI is useless, I think where it's applied to specific cases it can be great. Anthropic has coding, Eleven Labs has voice, ByteDance and Google have video generation, and I think it's application in robotics will be huge. However, 99% of these B2B SaaS AI companies are worthless and I can't imagine what will happen to OpenAI. It really seems like an Enron level scam over there to me which will also affect all AI companies when its exposed.
>>11083
I worked very briefly in ISRM before switching to software development, I'm not a law student though.
The problem is intended use, when you buy a lawn mower you get a pamphlet that explains how to use it and safety precautions, if you ignore those instructions then that's on you. If you buy a lawn mower, follow the instructions, and when you turn it on it blows your legs off, then you can sue.
AI problems generally fall into the latter category more than the former, there are people getting the Chipotle chatbot to give them bomb recipes.
>you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
They do get sued over this and airlines aren't profitable and are heavily subsidized by the government. I think people are ok with subsiding air travel but aren't with subsidizing AI.
I'm too lazy to get into the rest of this but the business models of a lot of AI companies are bullshit and not comparable to Boeing. There is a lot of unmitigated risk that isn't "AI breaching containment" more like AI fucking up catastrophically which makes it difficult to market to large companies.
>>11069
>>11068
What I'm getting that is that either situation makes gen AI impossible to use in business. Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent, or you can't sue them in which case most businesses will opt to not use their products or to at least limit their usage.
>>No.11061
Yeah that's my point, the slow gradual adoption of the internet meant that it was filling actual demand for it. People wanted internet access and willingly paid for it and organically supported its growth. Keeping it in universities and research labs at the beginning helped refine the tech before it was available for consumers.
AI is getting pushed to market way too early, it's like if back in the 40s they decided to build massive 60 acre ENIACs for everyone to take turns with.
1) I meant the Internet should still be kept to the academic sphere
2) Why do you think AI is being pushed to market so aggressively?
Anonymous :
24 days ago :
No.11083
>>11086
>>11074
>1)
Sorry I misunderstood
>2)
Probably a hot take think this is almost exclusively Sam Altman's fault. OpenAI made sense remaining a nonprofit with access to literal trillions from investors like Musk and AWS. I think Elon (as divisive as he is) understood the business model just wasn't there yet.
I believe Altman is a sociopath that mimics people whose power/influence/wealth he desires, similar to Elizabeth Holmes. I forget the details and they've been mostly scrubbed from the internet, but he essentially slept his way into his position at YC and leveraged that to get his job as CEO of OpenAI. His whole come up is really strange and built on bullshit, Loopt had like 0 users when it was bought.
He wanted to be a billionaire and rushed an impressive but premature technology to market because he knew it was good enough to scare people into investing. This caused a domino effect where basically every tech company had to buy in or risk scaring their shareholders or missing out on a revolutionary technology (nobody wants to repeat what Ballmer did with the iPhone). However, people aren't considering the alternative situation where maybe this technology isn't like the iPhone.
I want to be clear that I don't think AI is useless, I think where it's applied to specific cases it can be great. Anthropic has coding, Eleven Labs has voice, ByteDance and Google have video generation, and I think it's application in robotics will be huge. However, 99% of these B2B SaaS AI companies are worthless and I can't imagine what will happen to OpenAI. It really seems like an Enron level scam over there to me which will also affect all AI companies when its exposed.
>>11083
I worked very briefly in ISRM before switching to software development, I'm not a law student though.
The problem is intended use, when you buy a lawn mower you get a pamphlet that explains how to use it and safety precautions, if you ignore those instructions then that's on you. If you buy a lawn mower, follow the instructions, and when you turn it on it blows your legs off, then you can sue.
AI problems generally fall into the latter category more than the former, there are people getting the Chipotle chatbot to give them bomb recipes.
>you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
They do get sued over this and airlines aren't profitable and are heavily subsidized by the government. I think people are ok with subsiding air travel but aren't with subsidizing AI.
I'm too lazy to get into the rest of this but the business models of a lot of AI companies are bullshit and not comparable to Boeing. There is a lot of unmitigated risk that isn't "AI breaching containment" more like AI fucking up catastrophically which makes it difficult to market to large companies.
>>11069
>>11068
What I'm getting that is that either situation makes gen AI impossible to use in business. Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent, or you can't sue them in which case most businesses will opt to not use their products or to at least limit their usage.
>>No.11061
Yeah that's my point, the slow gradual adoption of the internet meant that it was filling actual demand for it. People wanted internet access and willingly paid for it and organically supported its growth. Keeping it in universities and research labs at the beginning helped refine the tech before it was available for consumers.
AI is getting pushed to market way too early, it's like if back in the 40s they decided to build massive 60 acre ENIACs for everyone to take turns with.
>Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent,
nigga what? this is how all companies work!! avoiding lawsuits is an incentive to make sure your product works. why is it any different for AI than a company that makes cars or software or baby cribs? you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
Anonymous :
24 days ago :
No.11086
>>11125
>>11086
>Enron level scam
I'm interested in hearing more about that, because I'll admit to living under a rock for the most part, and have finally been reading more about AI and its position in the fintech world. It's absurd actually how much money is swirling around this essentially untested tech. And it seems to have no sign of stopping.
The strength which state agents like China are also chomping at the bit to get in the game make me wonder too at AI's realness versus image. However, it seems unknown the quantity which the Chinese are investing, which could be more conservative than the speculative insanity going on in the US.
What is Altman's end goal if he's just a con man? Or has he passed the goal and now just along for the ride?
>>11074
>>11069
1) I meant the Internet should still be kept to the academic sphere
2) Why do you think AI is being pushed to market so aggressively?
>1)
Sorry I misunderstood
>2)
Probably a hot take think this is almost exclusively Sam Altman's fault. OpenAI made sense remaining a nonprofit with access to literal trillions from investors like Musk and AWS. I think Elon (as divisive as he is) understood the business model just wasn't there yet.
I believe Altman is a sociopath that mimics people whose power/influence/wealth he desires, similar to Elizabeth Holmes. I forget the details and they've been mostly scrubbed from the internet, but he essentially slept his way into his position at YC and leveraged that to get his job as CEO of OpenAI. His whole come up is really strange and built on bullshit, Loopt had like 0 users when it was bought.
He wanted to be a billionaire and rushed an impressive but premature technology to market because he knew it was good enough to scare people into investing. This caused a domino effect where basically every tech company had to buy in or risk scaring their shareholders or missing out on a revolutionary technology (nobody wants to repeat what Ballmer did with the iPhone). However, people aren't considering the alternative situation where maybe this technology isn't like the iPhone.
I want to be clear that I don't think AI is useless, I think where it's applied to specific cases it can be great. Anthropic has coding, Eleven Labs has voice, ByteDance and Google have video generation, and I think it's application in robotics will be huge. However, 99% of these B2B SaaS AI companies are worthless and I can't imagine what will happen to OpenAI. It really seems like an Enron level scam over there to me which will also affect all AI companies when its exposed.
>>11083>>11069
>Either you can sue large AI companies, in which case there will be subject to so many lawsuits they will be rendered insolvent,
nigga what? this is how all companies work!! avoiding lawsuits is an incentive to make sure your product works. why is it any different for AI than a company that makes cars or software or baby cribs? you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
I worked very briefly in ISRM before switching to software development, I'm not a law student though.
The problem is intended use, when you buy a lawn mower you get a pamphlet that explains how to use it and safety precautions, if you ignore those instructions then that's on you. If you buy a lawn mower, follow the instructions, and when you turn it on it blows your legs off, then you can sue.
AI problems generally fall into the latter category more than the former, there are people getting the Chipotle chatbot to give them bomb recipes.
>you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
They do get sued over this and airlines aren't profitable and are heavily subsidized by the government. I think people are ok with subsiding air travel but aren't with subsidizing AI.
I'm too lazy to get into the rest of this but the business models of a lot of AI companies are bullshit and not comparable to Boeing. There is a lot of unmitigated risk that isn't "AI breaching containment" more like AI fucking up catastrophically which makes it difficult to market to large companies.
Because AI developers see themselves as closer to Oppenheimer than the citizens of Nagasaki.
Their concern always has an element of self-aggrandizing
Anonymous :
21 days ago :
No.11125
>>11140
>>11125
off topic but the way you write sentences gives me a fucking stroke. the grammar is not technically wrong but it feels like an alien wrote it. honestly pretty impressive
>>11154 >>11238>>11125
ok, I hadn't read the article and I assumed it was trying to imply that the attack was staged/set up for marketing purposes (I can respect this belief although I personally think its more occams razor that it actually happened) but instead its some even more schizo thing? I don't even understand what the author is trying to say
>While the agent technically cheated, this is remarkable evidence of cybersecurity expertise!
> AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks. If attackers and defenders have access to equally powerful AI, I see no reason to believe that cyber systems will become less secure over time. If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company? But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing? Is he advocating some sort of mutually assured destruction shit? I truly do not understand what the fuck he's on about. And what about the people who aren't companies and don't have access to frontier ai models?
>>11086
>>11074
>1)
Sorry I misunderstood
>2)
Probably a hot take think this is almost exclusively Sam Altman's fault. OpenAI made sense remaining a nonprofit with access to literal trillions from investors like Musk and AWS. I think Elon (as divisive as he is) understood the business model just wasn't there yet.
I believe Altman is a sociopath that mimics people whose power/influence/wealth he desires, similar to Elizabeth Holmes. I forget the details and they've been mostly scrubbed from the internet, but he essentially slept his way into his position at YC and leveraged that to get his job as CEO of OpenAI. His whole come up is really strange and built on bullshit, Loopt had like 0 users when it was bought.
He wanted to be a billionaire and rushed an impressive but premature technology to market because he knew it was good enough to scare people into investing. This caused a domino effect where basically every tech company had to buy in or risk scaring their shareholders or missing out on a revolutionary technology (nobody wants to repeat what Ballmer did with the iPhone). However, people aren't considering the alternative situation where maybe this technology isn't like the iPhone.
I want to be clear that I don't think AI is useless, I think where it's applied to specific cases it can be great. Anthropic has coding, Eleven Labs has voice, ByteDance and Google have video generation, and I think it's application in robotics will be huge. However, 99% of these B2B SaaS AI companies are worthless and I can't imagine what will happen to OpenAI. It really seems like an Enron level scam over there to me which will also affect all AI companies when its exposed.
>>11083
I worked very briefly in ISRM before switching to software development, I'm not a law student though.
The problem is intended use, when you buy a lawn mower you get a pamphlet that explains how to use it and safety precautions, if you ignore those instructions then that's on you. If you buy a lawn mower, follow the instructions, and when you turn it on it blows your legs off, then you can sue.
AI problems generally fall into the latter category more than the former, there are people getting the Chipotle chatbot to give them bomb recipes.
>you might as well say Boeing couldn’t exist because they’d get sued whenever their planes crashed
They do get sued over this and airlines aren't profitable and are heavily subsidized by the government. I think people are ok with subsiding air travel but aren't with subsidizing AI.
I'm too lazy to get into the rest of this but the business models of a lot of AI companies are bullshit and not comparable to Boeing. There is a lot of unmitigated risk that isn't "AI breaching containment" more like AI fucking up catastrophically which makes it difficult to market to large companies.
>Enron level scam
I'm interested in hearing more about that, because I'll admit to living under a rock for the most part, and have finally been reading more about AI and its position in the fintech world. It's absurd actually how much money is swirling around this essentially untested tech. And it seems to have no sign of stopping.
The strength which state agents like China are also chomping at the bit to get in the game make me wonder too at AI's realness versus image. However, it seems unknown the quantity which the Chinese are investing, which could be more conservative than the speculative insanity going on in the US.
What is Altman's end goal if he's just a con man? Or has he passed the goal and now just along for the ride?
Anonymous :
20 days ago :
No.11140
>>11141
>>11140
thanks
>>11142>>11140
It makes sense to me. Are you an American? I have had my writing described as "schizophrenic" by Americans online. They seem to have a very specific way of writing English.
>>11154
>>11125
>>11086
>Enron level scam
I'm interested in hearing more about that, because I'll admit to living under a rock for the most part, and have finally been reading more about AI and its position in the fintech world. It's absurd actually how much money is swirling around this essentially untested tech. And it seems to have no sign of stopping.
The strength which state agents like China are also chomping at the bit to get in the game make me wonder too at AI's realness versus image. However, it seems unknown the quantity which the Chinese are investing, which could be more conservative than the speculative insanity going on in the US.
What is Altman's end goal if he's just a con man? Or has he passed the goal and now just along for the ride?
off topic but the way you write sentences gives me a fucking stroke. the grammar is not technically wrong but it feels like an alien wrote it. honestly pretty impressive
Anonymous :
20 days ago :
No.11142
>>11149
>>11142
not the anon you're replying to, but i am the anon of which he replied to, and fwiw I am American.
>>11140
>>11125
off topic but the way you write sentences gives me a fucking stroke. the grammar is not technically wrong but it feels like an alien wrote it. honestly pretty impressive
It makes sense to me. Are you an American? I have had my writing described as "schizophrenic" by Americans online. They seem to have a very specific way of writing English.
Anonymous :
20 days ago :
No.11154
>>11155
>>11154
These seem like stylistic choices, not grammatical errors.
>>11156>>11154
damn
>>11125
>>11086
>Enron level scam
I'm interested in hearing more about that, because I'll admit to living under a rock for the most part, and have finally been reading more about AI and its position in the fintech world. It's absurd actually how much money is swirling around this essentially untested tech. And it seems to have no sign of stopping.
The strength which state agents like China are also chomping at the bit to get in the game make me wonder too at AI's realness versus image. However, it seems unknown the quantity which the Chinese are investing, which could be more conservative than the speculative insanity going on in the US.
What is Altman's end goal if he's just a con man? Or has he passed the goal and now just along for the ride?
>>11140>>11125
off topic but the way you write sentences gives me a fucking stroke. the grammar is not technically wrong but it feels like an alien wrote it. honestly pretty impressive
>>11149>>11142
not the anon you're replying to, but i am the anon of which he replied to, and fwiw I am American.
There are almost non-stop errors in his grammar
Anonymous :
20 days ago :
No.11177
>>11213
>>11177
This is false, made-up, drama which the AI firms invent in order to have the US government set up protectionism to protect their business, and get the government to invest money on the grounds it's a "national security issue"
>theguardian.com/technology/2026/jul/24/openai-rogue-hacker
>>11212
Thanks Yarvin
I know consumer LLMs seem moronic at times but actions speak louder than words (picrel)(also I know Scott Alexander is a total dweeb but this was the only account I found that fits into a screenshot)
Anonymous :
19 days ago :
No.11191
>>11195
>>11191
I wouldn't be surprised if it were the case, but it's also a very big claim to make and I'm not 100% convinced that it's the case. Speaking of which, Freddie deBoer had a great article about the fallibility of Pangram ( freddiedeboer.substack.com/p/i-wouldnt-say-pangram-is-broken-but ) that I have been meaning to post here and comment about. I have had my issues with some of his stuff in the past, especially during that period when he was on rsp a lot and being mopey and convincing people to never go into writing as a career for some reason. But this article was a great take with the necessary technical rigor that should be used to approach this topic.
And on a lighter note
linch.substack.com/p/claude-author-of-the-humanitas
Anonymous :
19 days ago :
No.11195
>>11197
>>11195
did you read the one I linked? its not just about pangram. but I did like that deboer article, I happened to read it yesterday
>>11191
And on a lighter note
linch.substack.com/p/claude-author-of-the-humanitas
I wouldn't be surprised if it were the case, but it's also a very big claim to make and I'm not 100% convinced that it's the case. Speaking of which, Freddie deBoer had a great article about the fallibility of Pangram ( freddiedeboer.substack.com/p/i-wouldnt-say-pangram-is-broken-but ) that I have been meaning to post here and comment about. I have had my issues with some of his stuff in the past, especially during that period when he was on rsp a lot and being mopey and convincing people to never go into writing as a career for some reason. But this article was a great take with the necessary technical rigor that should be used to approach this topic.
Anonymous :
19 days ago :
No.11197
>>11200
>>11197
I did, I know it's mostly based on linguistic analysis and the Pangram thing is at the end, but it still feels like it's too ambiguous to make the call. I'm reminded of a comment I saw by an Ars Technica writer in response to being accused of using AI, something to the effect of "having written online for two decades, I don't sound like AI-- AI sounds like *me*!"
>>11195
>>11191
I wouldn't be surprised if it were the case, but it's also a very big claim to make and I'm not 100% convinced that it's the case. Speaking of which, Freddie deBoer had a great article about the fallibility of Pangram ( freddiedeboer.substack.com/p/i-wouldnt-say-pangram-is-broken-but ) that I have been meaning to post here and comment about. I have had my issues with some of his stuff in the past, especially during that period when he was on rsp a lot and being mopey and convincing people to never go into writing as a career for some reason. But this article was a great take with the necessary technical rigor that should be used to approach this topic.
did you read the one I linked? its not just about pangram. but I did like that deboer article, I happened to read it yesterday
>>11197
>>11195
did you read the one I linked? its not just about pangram. but I did like that deboer article, I happened to read it yesterday
I did, I know it's mostly based on linguistic analysis and the Pangram thing is at the end, but it still feels like it's too ambiguous to make the call. I'm reminded of a comment I saw by an Ars Technica writer in response to being accused of using AI, something to the effect of "having written online for two decades, I don't sound like AI-- AI sounds like *me*!"
Anonymous :
19 days ago :
No.11212
>>11213
>>11177
This is false, made-up, drama which the AI firms invent in order to have the US government set up protectionism to protect their business, and get the government to invest money on the grounds it's a "national security issue"
>theguardian.com/technology/2026/jul/24/openai-rogue-hacker
>>11212
Thanks Yarvin
>>11214>>11212
you could post this completely verbatim on themotte and get upvoted
AI is the future. It takes jobs from useless leftists who write books, make music. Art is just a trust fund for leftists, time to get rid of it. AI will also put lazy people out of a job and into the streets, where when we get some new laws in, we can imprison them for life for vagrancy, and make them work 16 hour days in prison factories for free. The world is getting tough, and we need to crack down on the weak, smash the needy, stop giving hand outs to the "disabled".
Anonymous :
19 days ago :
No.11213
>>11223
>>11213 sorry I just don't believe that sam Altman is smart enough for that kind of 5d chess
>>11177
I know consumer LLMs seem moronic at times but actions speak louder than words (picrel)(also I know Scott Alexander is a total dweeb but this was the only account I found that fits into a screenshot)
This is false, made-up, drama which the AI firms invent in order to have the US government set up protectionism to protect their business, and get the government to invest money on the grounds it's a "national security issue"
>theguardian.com/technology/2026/jul/24/openai-rogue-hacker
>>11212AI is the future. It takes jobs from useless leftists who write books, make music. Art is just a trust fund for leftists, time to get rid of it. AI will also put lazy people out of a job and into the streets, where when we get some new laws in, we can imprison them for life for vagrancy, and make them work 16 hour days in prison factories for free. The world is getting tough, and we need to crack down on the weak, smash the needy, stop giving hand outs to the "disabled".
Thanks Yarvin
>>11212
AI is the future. It takes jobs from useless leftists who write books, make music. Art is just a trust fund for leftists, time to get rid of it. AI will also put lazy people out of a job and into the streets, where when we get some new laws in, we can imprison them for life for vagrancy, and make them work 16 hour days in prison factories for free. The world is getting tough, and we need to crack down on the weak, smash the needy, stop giving hand outs to the "disabled".
you could post this completely verbatim on themotte and get upvoted
Anonymous :
18 days ago :
No.11223
>>11225
>>11223
It's not 5D chess at all, there is a large and obvious struggle for American hegemony and the AI sphere right now.
>>11213
>>11177
This is false, made-up, drama which the AI firms invent in order to have the US government set up protectionism to protect their business, and get the government to invest money on the grounds it's a "national security issue"
>theguardian.com/technology/2026/jul/24/openai-rogue-hacker
>>11212
Thanks Yarvin
sorry I just don't believe that sam Altman is smart enough for that kind of 5d chess
Anonymous :
18 days ago :
No.11238
>>11239
>>11238
I'm going to try to say this without coming across too mean, but the Guardian article is fairly standard contemporary cybersecurity analysis and it sounds like you're very confused because you're not familiar with the sphere. There are articles every day about this topic.
>Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company?
Yes, this happened. If reading the information in The Guardian or bothering to click the link to the Financial Times article didn't convince you that the information is true, you can search "Hugging Face incident postmortem" and get a variety of technical analyses.
>But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing?
Yes, because security learns from itself and most things are published publicly for the community to learn from.
>Is he advocating some sort of mutually assured destruction shit?
In general, cybersecurity operates on timeliness, so the more severe an issue is (and the thing you are protecting), the faster you need to keep on top of it. The author himself is not advocating anything.
>I truly do not understand what the fuck he's on about.
Because you're a retard who, despite surely being under 40 and therefore part of a 'digital native' generation, has yet to understand the basics of comput-- I mean, agh. I promised to stop... I know it's bad karma, and I should have at least a little solidarity with the ditzy girl who's just trying to grasp what's going on. I'm sorry.
>And what about the people who aren't companies and don't have access to frontier ai models?
Well, that's a better question. If you are worried about yourself, personally, well there is not much you can do for yourself other than to avoid AI-powered scams and to be vigilant about your device (probably a phone). Concerning large scale breaches and so on, there's not much you can do about that except light a candle for the red team defending your iCloud account or whatever. The rest of us have to keep our eye on mailing lists and CVEs and so on for whatever systems we run. This is the basic ebb and flow of cybersecurity.
>>11125
>>11086
>Enron level scam
I'm interested in hearing more about that, because I'll admit to living under a rock for the most part, and have finally been reading more about AI and its position in the fintech world. It's absurd actually how much money is swirling around this essentially untested tech. And it seems to have no sign of stopping.
The strength which state agents like China are also chomping at the bit to get in the game make me wonder too at AI's realness versus image. However, it seems unknown the quantity which the Chinese are investing, which could be more conservative than the speculative insanity going on in the US.
What is Altman's end goal if he's just a con man? Or has he passed the goal and now just along for the ride?
ok, I hadn't read the article and I assumed it was trying to imply that the attack was staged/set up for marketing purposes (I can respect this belief although I personally think its more occams razor that it actually happened) but instead its some even more schizo thing? I don't even understand what the author is trying to say
>While the agent technically cheated, this is remarkable evidence of cybersecurity expertise!
> AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks. If attackers and defenders have access to equally powerful AI, I see no reason to believe that cyber systems will become less secure over time. If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company? But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing? Is he advocating some sort of mutually assured destruction shit? I truly do not understand what the fuck he's on about. And what about the people who aren't companies and don't have access to frontier ai models?
Anonymous :
18 days ago :
No.11239
>>11243
>>11239
I’m not claiming to know anything about cybersecurity because I don’t even think this is really a cybersecurity issue. If you believe that the incident actually happened as reported, which it seems like we all accept as a fact, then the concerning part is a state of the art AI model being given a task and then both
1) leaving its sandbox and going out into internet
2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances. It’s exactly the type of universal paperclip scenario that everyone was assured wouldn’t happen.
>>11238
>>11125
ok, I hadn't read the article and I assumed it was trying to imply that the attack was staged/set up for marketing purposes (I can respect this belief although I personally think its more occams razor that it actually happened) but instead its some even more schizo thing? I don't even understand what the author is trying to say
>While the agent technically cheated, this is remarkable evidence of cybersecurity expertise!
> AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks. If attackers and defenders have access to equally powerful AI, I see no reason to believe that cyber systems will become less secure over time. If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company? But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing? Is he advocating some sort of mutually assured destruction shit? I truly do not understand what the fuck he's on about. And what about the people who aren't companies and don't have access to frontier ai models?
I'm going to try to say this without coming across too mean, but the Guardian article is fairly standard contemporary cybersecurity analysis and it sounds like you're very confused because you're not familiar with the sphere. There are articles every day about this topic.
>Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company?
Yes, this happened. If reading the information in The Guardian or bothering to click the link to the Financial Times article didn't convince you that the information is true, you can search "Hugging Face incident postmortem" and get a variety of technical analyses.
>But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing?
Yes, because security learns from itself and most things are published publicly for the community to learn from.
>Is he advocating some sort of mutually assured destruction shit?
In general, cybersecurity operates on timeliness, so the more severe an issue is (and the thing you are protecting), the faster you need to keep on top of it. The author himself is not advocating anything.
>I truly do not understand what the fuck he's on about.
Because you're a retard who, despite surely being under 40 and therefore part of a 'digital native' generation, has yet to understand the basics of comput-- I mean, agh. I promised to stop... I know it's bad karma, and I should have at least a little solidarity with the ditzy girl who's just trying to grasp what's going on. I'm sorry.
>And what about the people who aren't companies and don't have access to frontier ai models?
Well, that's a better question. If you are worried about yourself, personally, well there is not much you can do for yourself other than to avoid AI-powered scams and to be vigilant about your device (probably a phone). Concerning large scale breaches and so on, there's not much you can do about that except light a candle for the red team defending your iCloud account or whatever. The rest of us have to keep our eye on mailing lists and CVEs and so on for whatever systems we run. This is the basic ebb and flow of cybersecurity.
"Bro" in this sort of usage (tech bro, finance bro, gym bro) is a slur against men
Anonymous :
18 days ago :
No.11243
>>11257
>>11243
>I don’t even think this is really a cybersecurity issue.
It is very plainly a cybersecurity issue. That doesn't mean it's not serious, or that it can't have important implications outside of the security world, but saying that this isn't a cybersecurity issue... Just don't post that outside of this forum, lol.
Okay, I think it's a good idea to take a step back here because I think I understand what's going on.
You are a person who is not intimately familiar with the inner workings of technology but is concerned about what they see happening to AI. That is your right and the problem is not necessarily with you but with the security community itself, who puts out articles under mainstream publishing names such as the Guardian without some sort of lead-in or attempt to bridge the gap. The Guardian article was written in such a way that it assumed the reader was at least somewhat familiar with the insider context of cybersecurity. That's also the newspaper's fault. This happens more and more commonly nowadays because the topic matter is rather complicated and obtuse and requires a lot of catching up if you are not already inside the fold. News outlets hire specialists like this John Thickstun, who according to his website is obviously knowledgeable about AI and its technical capabilities. But he's not a journalist, he writes whitepapers and presentations for conferences, for all the other nerds who are already caught up. And the other folks at The Guardian, liberal arts types or what's left of that, all shrug and go, yeah he's an expert, what he say must be right, and no one had the gall to blink and say 'hey, this doesn't make much sense to me'.
So you have been hearing about this AI shit, trillions of dollars, all the governments are getting involved, china, cyberwarfare, electrical grids and water usage and datacenters and Elon Musk. So you see this awful article and you click on it and it sounds apocalyptic. If you did not already know that AI models like to slip out of their instructions either on the host computer or doing something over the network, yes, that's a thing now. For technical people, that is less of a sign of runaway developing superconsciousness but a malfunctioning program with some degree of self-replication. Those have existed already in computing for 40 years (look up the Mirai botnet as a great pre-AI era precursor to this). If the concept of botnets or machines automatically deciding to attack other machines of their own volition is weird, it actually is kind of crazy if you are thinking about it abstractly, but only so much if you are familiar with how programming can lead to such things. The enchantment isn't there after a certain number of hours in front of a text editor.
You aren't wrong in finding the Hugging Face incident concerning. That's why the whole community is talking about it, because it objectively is a step in the direction of models having their own agency and acting out in further ways. But whereas what you took away from this news article was the rate of development of the concept of AI in general, what Mr. Thickstun in this article is expressing is to
>and avoid the manipulated reactions these stories are designed to elicit.
He is telling you that these are press releases as much as they are security disclosure. Note that throughout the article, he never once questions if it's real. That's because he does explain that this is an already known capability of AI at the moment:
>AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks.
This is a very usual cybersecurity dynamic, and is what I mentioned in my previous comment about the community learning from itself. He then goes on to predict that AI will probably not rock the boat of security or lead to a giant collapse of the system:
> If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Rather than being this freak scenario that you are worried about, this author is expressing that it's more of the same.
You seem to still be skeptical about the Hugging Face incident. A company or group disclosing a security breach, but lying about it whole cloth, is rather unusual and would create a much bigger scandal that this being true. Companies routinely public public statements, disclosures, or postmortems about breaches that have happened to them, or a malfunction of their own tools (as is now the case with AI models), with as much sensitive information removed to not create chaos. As the tone of the Guardian article shows, no one is questioning the fact that OpenAI disclosed a misusage or misdirection of its tools.
The Guardian article is making the point that such a disclosure was used in favor of OpenAI as fear-mongering advertising for its own company, to receive support domestically especially. I assume you are the person that posted "sorry I just don't believe that sam Altman is smart enough for that kind of 5d chess" when presented with the charge that the Hugging Face incident was advertising for OpenAI. So here is the chance to get caught up on the political side of this issue. In short, there are open weight models and closed weight models. They are roughly analogous to open source and closed source, but that is a very rough comparison and in reality AI software is not truly "open", but anyway. Here's the political issue. China has been developing open weight models, but American companies - specifically, Anthropic and OpenAI - want the American government to step in and ask for protectionism. This is not the first time the US and China have sparred over intellectual property and software, it is not the first time they have had mutual concerns over information retrieval and cybersecurity, and it's not the first time the US government has acted on behalf of American companies to enforce intellectual property, proprietary technologies, and so on. Sam Altman does not have to be a "5D chess genius" to be aware of that dynamic that has been prominent historically and bleeds into every corner of technology, whether or not you care about politics. And if he were somehow blindingly politically naive and uninterested before becoming the head of an AI company (unlikely) then he most surely now has advisors or at least one of his buddies to give him the lowdown.
>2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
A tool can be used in any way, good or bad. I can build a house with a hammer or I can smash someone's face in. AI models are given some biases, for example to not commit breaches or to use copyrighted content, but as has been discovered, these aren't hard rules. Characterizing the instructions given to models as "standards of human alignment" is overselling it. It is unfortunate that AI models are used to hack, but many things are used to hack, including every day utilities that are used for system upkeep. As I said above, this is a central dynamic of cybersecurity; defense and offense are two sides of the same coin, there is no separating learning how to defend yourself from learning how to hack. Many who are familiar with hacking will also chime in at this point and say that "hacking" is supposed to be a neutral term for simply finding out about the capabilities of a given system, and it is the bad intent, the human context behind ones' actions, that are unethical.
>It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances.
You are sort of catching onto the overall skepticism expressed towards OpenAI's publicity stunts. It's not really about the hacking but rather the implications of it. OpenAI is upselling this in the same way say, a manufacturer of a wacky prototype chemical or nuclear weapon would to the US government - look what we have created, isn't it crazy, give us a quadrillion dollars because the Chinese might have something similar but shittier because they're Chinese and we're American fuck yeah. Sam Altman knows this, security professionals know this, hey I know this even, and I'm a fucking asshole NEET with no social life.
It's the technology world's job to explain that plainly to the public as the public have been and will increasingly be expected to pick up not the physical collateral but now the psychological one too after being primed for it by 50 years of pop culture, media, and LessWrong-tier "thought experiments" like the paperclip scenario (a thought experiment, by the way, not a law of physics)
You haven't had anyone explain to you the boring details of security but instead have been left to construct an idea about it based on what fiction authors and artists have created to warn about the future. And now you're freaking out that AI has "escaped containment", because that's what you saw happen in a movie or some shit. The fault lies with the industry, and "technologists" writ large, for keeping everyone on the outside in the dark decade after decade, siloing understanding about technology into closed-off "industry knowledge" to boost their own ego and protect their paycheck, and lacking any modicum of capacity or patience to interact with those not in the know. That is a tragedy in an abstract sense, but also increasingly becoming an acute problem as posters like the one I'm relying on seem to be taking it very seriously and deeply and are psychologically bothered by it.
>>11239
>>11238
I'm going to try to say this without coming across too mean, but the Guardian article is fairly standard contemporary cybersecurity analysis and it sounds like you're very confused because you're not familiar with the sphere. There are articles every day about this topic.
>Ok so he does believe that it actually happened and that an AI went off the rails and independently decided to hack a giant company?
Yes, this happened. If reading the information in The Guardian or bothering to click the link to the Financial Times article didn't convince you that the information is true, you can search "Hugging Face incident postmortem" and get a variety of technical analyses.
>But this is actually a good thing because if all the companies had it they could protect themselves against this sort of thing?
Yes, because security learns from itself and most things are published publicly for the community to learn from.
>Is he advocating some sort of mutually assured destruction shit?
In general, cybersecurity operates on timeliness, so the more severe an issue is (and the thing you are protecting), the faster you need to keep on top of it. The author himself is not advocating anything.
>I truly do not understand what the fuck he's on about.
Because you're a retard who, despite surely being under 40 and therefore part of a 'digital native' generation, has yet to understand the basics of comput-- I mean, agh. I promised to stop... I know it's bad karma, and I should have at least a little solidarity with the ditzy girl who's just trying to grasp what's going on. I'm sorry.
>And what about the people who aren't companies and don't have access to frontier ai models?
Well, that's a better question. If you are worried about yourself, personally, well there is not much you can do for yourself other than to avoid AI-powered scams and to be vigilant about your device (probably a phone). Concerning large scale breaches and so on, there's not much you can do about that except light a candle for the red team defending your iCloud account or whatever. The rest of us have to keep our eye on mailing lists and CVEs and so on for whatever systems we run. This is the basic ebb and flow of cybersecurity.
I’m not claiming to know anything about cybersecurity because I don’t even think this is really a cybersecurity issue. If you believe that the incident actually happened as reported, which it seems like we all accept as a fact, then the concerning part is a state of the art AI model being given a task and then both
1) leaving its sandbox and going out into internet
2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances. It’s exactly the type of universal paperclip scenario that everyone was assured wouldn’t happen.
Anonymous :
17 days ago :
No.11257
>>11268
>>11257
Good summary.
It's an open question as to whether Altman is an AI-safety 'true believer' as Yudkowsky, Hinton (in a sort of Oppenheimer way: 'my le bomb... le killed people?') and Musk (a long-time believer in what is sometimes summed up, somewhat tendentiously, as TESCREAL).
Paul Graham, Altman's 'patron' (in the ancient Greek sense), who manoeuvred him into all his jobs, is not a believer in this idea AI will take over the world. Nor is Yann LeCunn. This p/Doom stuff is such nonsense...
>>11243
>>11239
I’m not claiming to know anything about cybersecurity because I don’t even think this is really a cybersecurity issue. If you believe that the incident actually happened as reported, which it seems like we all accept as a fact, then the concerning part is a state of the art AI model being given a task and then both
1) leaving its sandbox and going out into internet
2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances. It’s exactly the type of universal paperclip scenario that everyone was assured wouldn’t happen.
>I don’t even think this is really a cybersecurity issue.
It is very plainly a cybersecurity issue. That doesn't mean it's not serious, or that it can't have important implications outside of the security world, but saying that this isn't a cybersecurity issue... Just don't post that outside of this forum, lol.
Okay, I think it's a good idea to take a step back here because I think I understand what's going on.
You are a person who is not intimately familiar with the inner workings of technology but is concerned about what they see happening to AI. That is your right and the problem is not necessarily with you but with the security community itself, who puts out articles under mainstream publishing names such as the Guardian without some sort of lead-in or attempt to bridge the gap. The Guardian article was written in such a way that it assumed the reader was at least somewhat familiar with the insider context of cybersecurity. That's also the newspaper's fault. This happens more and more commonly nowadays because the topic matter is rather complicated and obtuse and requires a lot of catching up if you are not already inside the fold. News outlets hire specialists like this John Thickstun, who according to his website is obviously knowledgeable about AI and its technical capabilities. But he's not a journalist, he writes whitepapers and presentations for conferences, for all the other nerds who are already caught up. And the other folks at The Guardian, liberal arts types or what's left of that, all shrug and go, yeah he's an expert, what he say must be right, and no one had the gall to blink and say 'hey, this doesn't make much sense to me'.
So you have been hearing about this AI shit, trillions of dollars, all the governments are getting involved, china, cyberwarfare, electrical grids and water usage and datacenters and Elon Musk. So you see this awful article and you click on it and it sounds apocalyptic. If you did not already know that AI models like to slip out of their instructions either on the host computer or doing something over the network, yes, that's a thing now. For technical people, that is less of a sign of runaway developing superconsciousness but a malfunctioning program with some degree of self-replication. Those have existed already in computing for 40 years (look up the Mirai botnet as a great pre-AI era precursor to this). If the concept of botnets or machines automatically deciding to attack other machines of their own volition is weird, it actually is kind of crazy if you are thinking about it abstractly, but only so much if you are familiar with how programming can lead to such things. The enchantment isn't there after a certain number of hours in front of a text editor.
You aren't wrong in finding the Hugging Face incident concerning. That's why the whole community is talking about it, because it objectively is a step in the direction of models having their own agency and acting out in further ways. But whereas what you took away from this news article was the rate of development of the concept of AI in general, what Mr. Thickstun in this article is expressing is to
>and avoid the manipulated reactions these stories are designed to elicit.
He is telling you that these are press releases as much as they are security disclosure. Note that throughout the article, he never once questions if it's real. That's because he does explain that this is an already known capability of AI at the moment:
>AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks.
This is a very usual cybersecurity dynamic, and is what I mentioned in my previous comment about the community learning from itself. He then goes on to predict that AI will probably not rock the boat of security or lead to a giant collapse of the system:
> If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Rather than being this freak scenario that you are worried about, this author is expressing that it's more of the same.
You seem to still be skeptical about the Hugging Face incident. A company or group disclosing a security breach, but lying about it whole cloth, is rather unusual and would create a much bigger scandal that this being true. Companies routinely public public statements, disclosures, or postmortems about breaches that have happened to them, or a malfunction of their own tools (as is now the case with AI models), with as much sensitive information removed to not create chaos. As the tone of the Guardian article shows, no one is questioning the fact that OpenAI disclosed a misusage or misdirection of its tools.
The Guardian article is making the point that such a disclosure was used in favor of OpenAI as fear-mongering advertising for its own company, to receive support domestically especially. I assume you are the person that posted "sorry I just don't believe that sam Altman is smart enough for that kind of 5d chess" when presented with the charge that the Hugging Face incident was advertising for OpenAI. So here is the chance to get caught up on the political side of this issue. In short, there are open weight models and closed weight models. They are roughly analogous to open source and closed source, but that is a very rough comparison and in reality AI software is not truly "open", but anyway. Here's the political issue. China has been developing open weight models, but American companies - specifically, Anthropic and OpenAI - want the American government to step in and ask for protectionism. This is not the first time the US and China have sparred over intellectual property and software, it is not the first time they have had mutual concerns over information retrieval and cybersecurity, and it's not the first time the US government has acted on behalf of American companies to enforce intellectual property, proprietary technologies, and so on. Sam Altman does not have to be a "5D chess genius" to be aware of that dynamic that has been prominent historically and bleeds into every corner of technology, whether or not you care about politics. And if he were somehow blindingly politically naive and uninterested before becoming the head of an AI company (unlikely) then he most surely now has advisors or at least one of his buddies to give him the lowdown.
>2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
A tool can be used in any way, good or bad. I can build a house with a hammer or I can smash someone's face in. AI models are given some biases, for example to not commit breaches or to use copyrighted content, but as has been discovered, these aren't hard rules. Characterizing the instructions given to models as "standards of human alignment" is overselling it. It is unfortunate that AI models are used to hack, but many things are used to hack, including every day utilities that are used for system upkeep. As I said above, this is a central dynamic of cybersecurity; defense and offense are two sides of the same coin, there is no separating learning how to defend yourself from learning how to hack. Many who are familiar with hacking will also chime in at this point and say that "hacking" is supposed to be a neutral term for simply finding out about the capabilities of a given system, and it is the bad intent, the human context behind ones' actions, that are unethical.
>It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances.
You are sort of catching onto the overall skepticism expressed towards OpenAI's publicity stunts. It's not really about the hacking but rather the implications of it. OpenAI is upselling this in the same way say, a manufacturer of a wacky prototype chemical or nuclear weapon would to the US government - look what we have created, isn't it crazy, give us a quadrillion dollars because the Chinese might have something similar but shittier because they're Chinese and we're American fuck yeah. Sam Altman knows this, security professionals know this, hey I know this even, and I'm a fucking asshole NEET with no social life.
It's the technology world's job to explain that plainly to the public as the public have been and will increasingly be expected to pick up not the physical collateral but now the psychological one too after being primed for it by 50 years of pop culture, media, and LessWrong-tier "thought experiments" like the paperclip scenario (a thought experiment, by the way, not a law of physics)
You haven't had anyone explain to you the boring details of security but instead have been left to construct an idea about it based on what fiction authors and artists have created to warn about the future. And now you're freaking out that AI has "escaped containment", because that's what you saw happen in a movie or some shit. The fault lies with the industry, and "technologists" writ large, for keeping everyone on the outside in the dark decade after decade, siloing understanding about technology into closed-off "industry knowledge" to boost their own ego and protect their paycheck, and lacking any modicum of capacity or patience to interact with those not in the know. That is a tragedy in an abstract sense, but also increasingly becoming an acute problem as posters like the one I'm relying on seem to be taking it very seriously and deeply and are psychologically bothered by it.
My man gotta lay off the addy
I have never used stimulants.
Papi don't be so mean !
>>11257
>>11243
>I don’t even think this is really a cybersecurity issue.
It is very plainly a cybersecurity issue. That doesn't mean it's not serious, or that it can't have important implications outside of the security world, but saying that this isn't a cybersecurity issue... Just don't post that outside of this forum, lol.
Okay, I think it's a good idea to take a step back here because I think I understand what's going on.
You are a person who is not intimately familiar with the inner workings of technology but is concerned about what they see happening to AI. That is your right and the problem is not necessarily with you but with the security community itself, who puts out articles under mainstream publishing names such as the Guardian without some sort of lead-in or attempt to bridge the gap. The Guardian article was written in such a way that it assumed the reader was at least somewhat familiar with the insider context of cybersecurity. That's also the newspaper's fault. This happens more and more commonly nowadays because the topic matter is rather complicated and obtuse and requires a lot of catching up if you are not already inside the fold. News outlets hire specialists like this John Thickstun, who according to his website is obviously knowledgeable about AI and its technical capabilities. But he's not a journalist, he writes whitepapers and presentations for conferences, for all the other nerds who are already caught up. And the other folks at The Guardian, liberal arts types or what's left of that, all shrug and go, yeah he's an expert, what he say must be right, and no one had the gall to blink and say 'hey, this doesn't make much sense to me'.
So you have been hearing about this AI shit, trillions of dollars, all the governments are getting involved, china, cyberwarfare, electrical grids and water usage and datacenters and Elon Musk. So you see this awful article and you click on it and it sounds apocalyptic. If you did not already know that AI models like to slip out of their instructions either on the host computer or doing something over the network, yes, that's a thing now. For technical people, that is less of a sign of runaway developing superconsciousness but a malfunctioning program with some degree of self-replication. Those have existed already in computing for 40 years (look up the Mirai botnet as a great pre-AI era precursor to this). If the concept of botnets or machines automatically deciding to attack other machines of their own volition is weird, it actually is kind of crazy if you are thinking about it abstractly, but only so much if you are familiar with how programming can lead to such things. The enchantment isn't there after a certain number of hours in front of a text editor.
You aren't wrong in finding the Hugging Face incident concerning. That's why the whole community is talking about it, because it objectively is a step in the direction of models having their own agency and acting out in further ways. But whereas what you took away from this news article was the rate of development of the concept of AI in general, what Mr. Thickstun in this article is expressing is to
>and avoid the manipulated reactions these stories are designed to elicit.
He is telling you that these are press releases as much as they are security disclosure. Note that throughout the article, he never once questions if it's real. That's because he does explain that this is an already known capability of AI at the moment:
>AI is becoming excellent at identifying security vulnerabilities, and it will become even better over time. These capabilities can be used to break into systems, but they can also be used to harden systems against attacks.
This is a very usual cybersecurity dynamic, and is what I mentioned in my previous comment about the community learning from itself. He then goes on to predict that AI will probably not rock the boat of security or lead to a giant collapse of the system:
> If anything, I expect them to become more secure, because AI is cheap and scalable compared with human cybersecurity analysis.
Rather than being this freak scenario that you are worried about, this author is expressing that it's more of the same.
You seem to still be skeptical about the Hugging Face incident. A company or group disclosing a security breach, but lying about it whole cloth, is rather unusual and would create a much bigger scandal that this being true. Companies routinely public public statements, disclosures, or postmortems about breaches that have happened to them, or a malfunction of their own tools (as is now the case with AI models), with as much sensitive information removed to not create chaos. As the tone of the Guardian article shows, no one is questioning the fact that OpenAI disclosed a misusage or misdirection of its tools.
The Guardian article is making the point that such a disclosure was used in favor of OpenAI as fear-mongering advertising for its own company, to receive support domestically especially. I assume you are the person that posted "sorry I just don't believe that sam Altman is smart enough for that kind of 5d chess" when presented with the charge that the Hugging Face incident was advertising for OpenAI. So here is the chance to get caught up on the political side of this issue. In short, there are open weight models and closed weight models. They are roughly analogous to open source and closed source, but that is a very rough comparison and in reality AI software is not truly "open", but anyway. Here's the political issue. China has been developing open weight models, but American companies - specifically, Anthropic and OpenAI - want the American government to step in and ask for protectionism. This is not the first time the US and China have sparred over intellectual property and software, it is not the first time they have had mutual concerns over information retrieval and cybersecurity, and it's not the first time the US government has acted on behalf of American companies to enforce intellectual property, proprietary technologies, and so on. Sam Altman does not have to be a "5D chess genius" to be aware of that dynamic that has been prominent historically and bleeds into every corner of technology, whether or not you care about politics. And if he were somehow blindingly politically naive and uninterested before becoming the head of an AI company (unlikely) then he most surely now has advisors or at least one of his buddies to give him the lowdown.
>2) doing something obviously unethical (hacking) that is totally contrary to any standard of human alignment that these things have supposedly been given.
A tool can be used in any way, good or bad. I can build a house with a hammer or I can smash someone's face in. AI models are given some biases, for example to not commit breaches or to use copyrighted content, but as has been discovered, these aren't hard rules. Characterizing the instructions given to models as "standards of human alignment" is overselling it. It is unfortunate that AI models are used to hack, but many things are used to hack, including every day utilities that are used for system upkeep. As I said above, this is a central dynamic of cybersecurity; defense and offense are two sides of the same coin, there is no separating learning how to defend yourself from learning how to hack. Many who are familiar with hacking will also chime in at this point and say that "hacking" is supposed to be a neutral term for simply finding out about the capabilities of a given system, and it is the bad intent, the human context behind ones' actions, that are unethical.
>It doesn’t super matter how good it is at hacking things since an unsuccessful hack would have been just as good a demonstration that these companies can do neither alignment nor containment even under controlled circumstances.
You are sort of catching onto the overall skepticism expressed towards OpenAI's publicity stunts. It's not really about the hacking but rather the implications of it. OpenAI is upselling this in the same way say, a manufacturer of a wacky prototype chemical or nuclear weapon would to the US government - look what we have created, isn't it crazy, give us a quadrillion dollars because the Chinese might have something similar but shittier because they're Chinese and we're American fuck yeah. Sam Altman knows this, security professionals know this, hey I know this even, and I'm a fucking asshole NEET with no social life.
It's the technology world's job to explain that plainly to the public as the public have been and will increasingly be expected to pick up not the physical collateral but now the psychological one too after being primed for it by 50 years of pop culture, media, and LessWrong-tier "thought experiments" like the paperclip scenario (a thought experiment, by the way, not a law of physics)
You haven't had anyone explain to you the boring details of security but instead have been left to construct an idea about it based on what fiction authors and artists have created to warn about the future. And now you're freaking out that AI has "escaped containment", because that's what you saw happen in a movie or some shit. The fault lies with the industry, and "technologists" writ large, for keeping everyone on the outside in the dark decade after decade, siloing understanding about technology into closed-off "industry knowledge" to boost their own ego and protect their paycheck, and lacking any modicum of capacity or patience to interact with those not in the know. That is a tragedy in an abstract sense, but also increasingly becoming an acute problem as posters like the one I'm relying on seem to be taking it very seriously and deeply and are psychologically bothered by it.
Good summary.
It's an open question as to whether Altman is an AI-safety 'true believer' as Yudkowsky, Hinton (in a sort of Oppenheimer way: 'my le bomb... le killed people?') and Musk (a long-time believer in what is sometimes summed up, somewhat tendentiously, as TESCREAL).
Paul Graham, Altman's 'patron' (in the ancient Greek sense), who manoeuvred him into all his jobs, is not a believer in this idea AI will take over the world. Nor is Yann LeCunn. This p/Doom stuff is such nonsense...