Absolutely true in medicine. Tech has “replaced” 90% of the work that transcriptionists, scribes, medical coder/billers and after hours answering services did 20 years ago. Guess who usually is expected to pick up the 10% of those jobs that tech can’t quite do yet…the docs. Despite our highly specialized training, this workflow often makes sense to the employer because docs are not paid hourly wages. The loss of productivity to the healthcare system is significant. Not to mention the epidemic of burnout among physicians who now spend a significant part of their day, usually after normal office hours, editing notes, tidying up coding/billing, and triaging after hours calls that AI can’t quite handle yet.
Transitional hybrid workflows will be needed until the tech gets closer to 99% human replacement performance. The hidden “cost” of early automation needs to be taken more seriously, at least in my field.
Your point about AI adoption in business is, I think, well taken. There seems to be a naive assumption made by many technologists, which is something like the following: because there is rapid improvement in AI tech, businesses will rapidly adopt the tech. But the second claim doesn't follow from the first; businesses are risk-averse and institutional inertia is real. To some extent the average technologist views AI as a kind of generalizable plug and play technology akin to electricity: just hook a business into AI, and, well, magic happens. Of course reality is much more complicated than that.
Absolutely true in medicine. Tech has “replaced” 90% of the work that transcriptionists, scribes, medical coder/billers and after hours answering services did 20 years ago. Guess who usually is expected to pick up the 10% of those jobs that tech can’t quite do yet…the docs. Despite our highly specialized training, this workflow often makes sense to the employer because docs are not paid hourly wages. The loss of productivity to the healthcare system is significant. Not to mention the epidemic of burnout among physicians who now spend a significant part of their day, usually after normal office hours, editing notes, tidying up coding/billing, and triaging after hours calls that AI can’t quite handle yet.
Transitional hybrid workflows will be needed until the tech gets closer to 99% human replacement performance. The hidden “cost” of early automation needs to be taken more seriously, at least in my field.
Solow pointed out that he saw computers everywhere but in the productivity statistics
Your point about AI adoption in business is, I think, well taken. There seems to be a naive assumption made by many technologists, which is something like the following: because there is rapid improvement in AI tech, businesses will rapidly adopt the tech. But the second claim doesn't follow from the first; businesses are risk-averse and institutional inertia is real. To some extent the average technologist views AI as a kind of generalizable plug and play technology akin to electricity: just hook a business into AI, and, well, magic happens. Of course reality is much more complicated than that.