The insidious part missing here is that AI search destroys the internet. You no longer search for other people’s pages or content … you simply search for “the answer”.
Sure there might be links and footnotes, but the whole product is to reconstitute the internet into something Google (or whoever) own and control from top to bottom. That is the death of the internet and some of the values which built it in the first place.
Ideally for Google, we all become “information or content” serfs to their AI “freehold”. Every “conversation” we have with the AI or otherwise is more training data. Every post or article or report or paper is just data for the AI which we provide as service to suckle at the great “AI search”.
And lets not fool ourselves into thinking that there isn’t real and convincing convenience in something like this. It makes sense, so long as AI can be useful enough to justify the easiness of it.
Which is why the real issue isn’t whether AI is “good enough” or “not actually intelligent” … that’s a distraction. The issue is what are the economic implications.
AI is hard to train and to keep up to date and it’s hard to improve on … these are resource intensive tasks. Which means there’s centralisation built right in.
AI consumes and stores data in a destructive way. It destroys or undermines the utility of that data … as you can just use the AI instead … and it is also likely lossy (thus hallucinations etc).
So … centralised data eating technology. If we were talking about liberties or property rights or IP or creativity or the economy … an all eating centralising pattern would be thunderously fearsome. Monarchism, Imperialism or colonialism … monopolisation … complete serfdom. It’s the same type of thing … but “just” for information technology … which is maybe not that significant … except how much are we all using the internet for anything and how much are our livelihoods linked to it in someway?
Very well said. I mean I think the internet is kinda garbage to begin with due to all the advertising and SEO junk that makes everything unreliable in the first place. We wouldn’t need AI in the first place if we could just find the information we need.
Also there’s a fair bit of effort going in to making local LLMs. I haven’t seen that they are as good, but they might be good enough. But at any rate, there isn’t any incentive to put knowledge on the internet any more because there’s no fame and glory or money in being part of a massive data set.
We wouldn’t need AI in the first place if we could just find the information we need.
I’ve commented on this before … but it almost seems like that is the point, or an opportunistic moment for Google … turn the internet to shit so that we “need” the AI and that Google have a new business to grow into. Capitalism at its finest.
Also there’s a fair bit of effort going in to making local LLMs.
Yea it’s definitely interesting but my gut feeling is that open source or local LLMs (like llama) are false hope against the broader dynamics. Surely with greater Google-level resources comes ‘better’ and more convenient AI. I’d bet that open/local LLMs will end up like Linux Desktop: meaninglessly small technical user base with no anti-monopoly effects at all. Which, to open the issue up to “capitalism!!”, raises the general issue of how individualistic rather than organisational actions can be ineffectual.
The insidious part missing here is that AI search destroys the internet. You no longer search for other people’s pages or content … you simply search for “the answer”.
Sure there might be links and footnotes, but the whole product is to reconstitute the internet into something Google (or whoever) own and control from top to bottom. That is the death of the internet and some of the values which built it in the first place.
Ideally for Google, we all become “information or content” serfs to their AI “freehold”. Every “conversation” we have with the AI or otherwise is more training data. Every post or article or report or paper is just data for the AI which we provide as service to suckle at the great “AI search”.
And lets not fool ourselves into thinking that there isn’t real and convincing convenience in something like this. It makes sense, so long as AI can be useful enough to justify the easiness of it.
Which is why the real issue isn’t whether AI is “good enough” or “not actually intelligent” … that’s a distraction. The issue is what are the economic implications.
AI is hard to train and to keep up to date and it’s hard to improve on … these are resource intensive tasks. Which means there’s centralisation built right in.
AI consumes and stores data in a destructive way. It destroys or undermines the utility of that data … as you can just use the AI instead … and it is also likely lossy (thus hallucinations etc).
So … centralised data eating technology. If we were talking about liberties or property rights or IP or creativity or the economy … an all eating centralising pattern would be thunderously fearsome. Monarchism, Imperialism or colonialism … monopolisation … complete serfdom. It’s the same type of thing … but “just” for information technology … which is maybe not that significant … except how much are we all using the internet for anything and how much are our livelihoods linked to it in someway?
Very well said. I mean I think the internet is kinda garbage to begin with due to all the advertising and SEO junk that makes everything unreliable in the first place. We wouldn’t need AI in the first place if we could just find the information we need.
Also there’s a fair bit of effort going in to making local LLMs. I haven’t seen that they are as good, but they might be good enough. But at any rate, there isn’t any incentive to put knowledge on the internet any more because there’s no fame and glory or money in being part of a massive data set.
I’ve commented on this before … but it almost seems like that is the point, or an opportunistic moment for Google … turn the internet to shit so that we “need” the AI and that Google have a new business to grow into. Capitalism at its finest.
Yea it’s definitely interesting but my gut feeling is that open source or local LLMs (like llama) are false hope against the broader dynamics. Surely with greater Google-level resources comes ‘better’ and more convenient AI. I’d bet that open/local LLMs will end up like Linux Desktop: meaninglessly small technical user base with no anti-monopoly effects at all. Which, to open the issue up to “capitalism!!”, raises the general issue of how individualistic rather than organisational actions can be ineffectual.