I believe AI is quite capable in the right circumstances, but I'm not convinced "this" is the watershed moment.
I believe AI is quite capable in the right circumstances, but I'm not convinced "this" is the watershed moment.
The right way to look at it is that LLMs help us to maximize the utility of the entirety of current human achievement/knowledge by discovering obscure connections in it.
Imagine SQL but the queries mostly write themselves and database is just entirety of human knowledge with no formalization. If you were to create such thing 7 years ago you'd say somebody expects a miracle out of you. I don't get why so many people have trouble recognizing it now as such.
Because AIs are sort of like that. Creating views of existing data no human has ever came up with. Views that might unlock new even more surprising and useful views. They are sort of materialized views because we are writing down what LLMs came up with. So they are new data tables. Derived, true. But possibly novel and unlocking something else.
In the case of LLM, a human has to create a "prompt" that the LLM just continue, based on the patterns humans has created in the past.