It won't kill llms, but it'll make curating the training data the most expensive and laborious part. No more freebies like feeding the entire internet into it
If LLMs heighten the contradictions of the political economy of science and push it towards a new model, I won't be upset about it.
In the current system, that only happens if can get immediately richer or if he can be convinced that it'll make him richer soon. The contradictions, like in so many other things, are otherwise maintained even at the expense of future profitability.
Bit idea: Use AI to generate your slop and get that published so AI uses AI generated slop to learn and continuously gets worse
a human centipede ouroboros, a machine shitting into its own mouth forever
Yes that's already being done and while it will kill LLMs, it's not very good for people who read or do science.
Yeah yeah they can be useful tools for specific tasks, but the hype and outright techbro occult euphoria needs to fucking die already.
https://futurism.com/openai-employees-say-firms-chief-scientist-has-been-making-strange-spiritual-claims
It won't kill llms, but it'll make curating the training data the most expensive and laborious part. No more freebies like feeding the entire internet into it
The field really needs to die and be reborn. The journal model is a relic of a time when only 200 million people could read and write.
Publication output has gone from 1 million in 2000 to 2 million in ~2010, with 3.3 million publications produced in 2022.
If LLMs heighten the contradictions of the political economy of science and push it towards a new model, I won't be upset about it.
In the current system, that only happens if
can get immediately richer or if he can be convinced that it'll make him richer soon. The contradictions, like in so many other things, are otherwise maintained even at the expense of future profitability.
In this paper AI will show . . .
I hope this answer was helpful!