Thanks. I have been looking forward to this. Tao's perspective is sadly absent from a lot of the debates I've seen surrounding this, even the less dogmatic and more substantial discussions. The greatest benefits of research and theory development are realized in the process, not the results. It's to our own great harm that we relegate these processes to the machines. That's not meant to suggest ignorance, only that these new tools will need to be leveraged responsibly. For those who espouse "self regulation", I believe this is one space that's appropriate and accessible for LLM providers, and I think it's sad that their first tests don't seem to suggest good faith engagement. It's never too late to do better.
Thanks. I have been looking forward to this. Tao's perspective is sadly absent from a lot of the debates I've seen surrounding this, even the less dogmatic and more substantial discussions. The greatest benefits of research and theory development are realized in the process, not the results. It's to our own great harm that we relegate these processes to the machines. That's not meant to suggest ignorance, only that these new tools will need to be leveraged responsibly. For those who espouse "self regulation", I believe this is one space that's appropriate and accessible for LLM providers, and I think it's sad that their first tests don't seem to suggest good faith engagement. It's never too late to do better.
Related: Two Dire Warnings, One From Terrance Tao - https://news.ycombinator.com/item?id=49620412