JenniferplusplusYou can't put the genie back in the bottle 🤡
Meanwhile, the genie is only kept out of the bottle by the force of literally the entirety of the world's available investment capital
Obviously, the models are going to continue to exist. Probably. So will the data sets, the libraries, and the generalized knowledge of how to build them and what that gets you, in a technical sense.
And if you think that's what AI is?
🤡
For the record, AI is a technocratic political project for the purpose of industrializing knowledge work.
That industrialization of knowledge work has 2 parts. First, is mechanization. Making knowledge work dependent on some particular machine. This is already the case, with computers generally. Knowledge work is thoroughly computerized. But, those computers are small, cheap, universally available commodities. That doesn't serve the second part, so they're forcing in new layers of mechanization, and removing access to the old machines.
Second, they ensure those machines can only be obtained through large investments of capital. Thus, all knowledge work can be done only at the pleasure of the capitalists who own the machines. Personal computers don't help them, there. But a black box hosted service that consumes the entire web to build and a whole country's worth of electricity to operate sure as hell does.
"AI" is merely the banner under which they are organizing and justifying this project. The implementation details are just the implementation details.
So you can see how that genie is extremely prone to returning to it's bottle.
It only stays out as long as they can keep shoveling an ever increasing amount of real resources into it. And it turns out the resources available are finite
Marc Trius@jenniferplusplus how do small language models that can be run locally or self hosted factor into this? Perhaps they represent tools that can still be controlled locally rather than by large investors.
Counter arguments:
1. They still need to be trained
2. Hardware prices are rising, hardware capable of running even small models may become out of reach
@alter_kaker mostly by recruiting people who should know better to defend the AI project, it seems
@jenniferplusplus I'm sorry, I don't understand
@alter_kaker how do small language models that can be run locally or self hosted factor into this?
They don't change anything at all about the power dynamics. They don't change anything at all about whether you can do or economically benefit from work that capitalists disapprove of. What they do is distract people who are concerned about those facts, and make everyone else spend time explaining to them that they don't actually change the political situation at all.