But the reality is that AI is being used in all sorts of professional environments in very productive ways now,
The reality is that that claim is anecdotal at best as there is no way to quantify the productivity of LLM use.
Productivity is value generated per unit of cost. You cannot reliably predict the token use of successfully completing any given task or even reproduce the same token use for running the same task again later. If you cannot quantify the cost you cannot quantify productivity.
Yes, my information is anecdotal, but that’s no reason to dismiss it out of hand. I personally know attorneys, physicians, and software engineers for whom AI has become an integral tool in their work. These are all super competent people who understand how to use it and never rely on it for a final draft. That it’s increased their productivity individually is frankly undeniable. It sounds like you’re talking about expecting AI to spit out a reliable final product, which is absolutely not what it should be used for
The reality is that that claim is anecdotal at best as there is no way to quantify the productivity of LLM use.
Productivity is value generated per unit of cost. You cannot reliably predict the token use of successfully completing any given task or even reproduce the same token use for running the same task again later. If you cannot quantify the cost you cannot quantify productivity.
Yes, my information is anecdotal, but that’s no reason to dismiss it out of hand. I personally know attorneys, physicians, and software engineers for whom AI has become an integral tool in their work. These are all super competent people who understand how to use it and never rely on it for a final draft. That it’s increased their productivity individually is frankly undeniable. It sounds like you’re talking about expecting AI to spit out a reliable final product, which is absolutely not what it should be used for