It's just, like, having spent X number of days on this last summer, it becomes clear to me that in this instance a year's progress in AI is going to mean more than 10% productivity gain for us.
On this one task! I don't think that generalizes to growing wheat and restocking shelves. But still.
On this one task! I don't think that generalizes to growing wheat and restocking shelves. But still.
Comments
Deployed along the lines of what it can do well -which necessarily includes "used by whom" and "used for what"- it does have some significant...
But in practice the deployment of LLM technologies --at the self-reinforcing intersection of incentives and fantasies-- has had so far a negative impact on the net, I think; enabled not the least by how good they are at something similar to but not the right thing.
The number of people who overrate LLM output so long as it’s distant from their own specialism is so depressing.