Text
The richest man on earth owns X.
The second richest man on earth is about to be a major owner of TikTok.
The third richest man owns Facebook, Instagram, and WhatsApp.
The fourth richest man owns The Washington Post.
See the problem here?
The richest man on earth owns X.
The second richest man on earth is about to be a major owner of TikTok.
The third richest man owns Facebook, Instagram, and WhatsApp.
The fourth richest man owns The Washington Post.
See the problem here?
what do LLMs have to do with this? /genq
They can (and will) be used to subtly manipulate facts and feelings, and to rewrite history if left unchallenged. We’re moving towards a future where access to information will be limited, not unlike ingsoc.
Make offline copies of Wikipedia and distribute them everywhere you can
Thats my weekend project and it takes a lot less space than you think (at least to me).
If you were born before the 90s, this is kind of how it was. The rich controlled tv, newspapers, magazines, etc. The only people who had access were the people who had access to great libraries for the old stuff. If you wanted breaking news, that was controlled.
Except it was a lot more rich people, most nowhere near as wealthy as those today. The consolidation which has been allowed since then is absurd.
True, there was a lot more diversity in what the greeds wanted said.
ohhhh okay yes i understand
that is why we need more open source LLMs!
Open source LLMs are mostly just open weights, so you still don’t know how biased the data was that they were trained on.
I suppose you could distribute the source code for a model and then distribute it with a whole bunch of training data that the end user can ingest on their own. But then that training data would have to be public domain.
Yes yes i do know my comment was talking about actually open sourcing LLMs, not about open weight LLMs
However, open weight LLMs are still infinitely better than using something like ChatGPT on OpenAI’s servers