Over the past month, my opinions on Anthropic have drastically shifted. I’ve had 4 pro claude subscriptions and still have 2 currently.
Google has released Gemma for open source models, OpenAI has released open source, xAI has released open source, Meta has released open source, literally every big AI lab has released open sourced models…
Except Anthropic.
Anthropic has not only not released a single open source model, they have for several years now been champions of the idea that open source AI is a bad thing and that’s it’s dangerous. The CEO uses mechanistic interpretability as the reason…which is so dumb. You can’t see inside of a model regardless, no one knows what happens inside of models. We know what happens but don’t know why. This is an open question in the field, including an entire division at Anthropic.
Imagine a world where the company with the largest and best AI in the world is so far ahead of everyone else that any other AI isn’t even basically an option, they only charge API pricing, and open source is banned.
That’s a dark Orwellian fantasy and if they have their way, that’s where we will be.
[link] [comments]