The head of "Safety" at Open AI
The head of "Safety" at Open AI
This is almost too sad to dunk on. People running the world have never had a reflective conversation with a professional trained to help process emotions and healthily cope with the human condition. They're in charge of billions of dollars worth of resources and guiding society's investment into the future. Cooooool
You're viewing a single thread.
Don't put it past her to just be completely lying and trying to sell a product.
Permanently Deleted
I see. A twofer.
From following machine learning for the past year, mostly focusing on open source development, this is almost certainly correct.
Most people don't realize this, but current conditions are actually against large AI companies. The only thing they can do that can't be done by anyone with a GPU made in the past 5 years is make larger models which aren't necessarily better and which are much slower to train and iterate on. Meanwhile, open source models have hordes of researchers writing papers and new breakthroughs come so fast people don't have time to widely implement them all. Google privately acknowledged this some time ago and noted that OpenAI is in the same position. They have no moat.
Pretty much the only way for OpenAI to maintain an advantage is to convince the public that their closed-source AI is safe, and that open source AI is too dangerous to allow, when in reality they are fundamentally the same. Granted, banning or regulation of open source AI products is not going to ever fully work, but they would be hoping that it slows it down enough.
What are some FOSS large language model alternatives? i'm curious to play with them.
There's mistral I guess, it's apache-licensed
well the underlying technology of chatGPT is Free and open source. They've just shoveled more data than usual into it