Even in software development, which the models are better trained for than other domains and where results are easily verifiable by tests and compilers, I haven't heard of much outright replacement. Maybe the businesses that have tried it are fumbling anyway and that's why they jumped on an unproven idea.<p>I think at the end of the day there will be a tremendous skills amplification in many fields but the amount of work to be done will expand with the lowered cost and the amount of jobs will decrease less than the doomsayers are predicting.