Except… science has shown this to be true. Even after a year-plus of work, less than 2% of devs work faster or more efficiently with AI than without it. And for almost 90% of the remainder, regression analysis strongly indicated that none of them would ever be better with AI than without it, regardless of how much practice they had with it.
These general results have come up with study after study over the last few years, with very consistent patterns. And with AI becoming more hallucinatory and downright wrong with every generation - about 60-80% of all responses with the latest models, depending on model being examined - the proportion of devs being able to wrestle AI into creating functionally viable work faster than they could to it themselves has also decreased slightly.
Except… science has shown this to be true. Even after a year-plus of work, less than 2% of devs work faster or more efficiently with AI than without it. And for almost 90% of the remainder, regression analysis strongly indicated that none of them would ever be better with AI than without it, regardless of how much practice they had with it.
These general results have come up with study after study over the last few years, with very consistent patterns. And with AI becoming more hallucinatory and downright wrong with every generation - about 60-80% of all responses with the latest models, depending on model being examined - the proportion of devs being able to wrestle AI into creating functionally viable work faster than they could to it themselves has also decreased slightly.