I think the best first target is augmenting specialists, not replacing them. But the problems are cultural much more than technical. Recent improvements in alg

I think the best first target is augmenting specialists, not replacing them. But the problems are cultural much more than technical. Recent improvements in algorithmic performance shouldn't blind people to the fact that for decades now ML approaches have been able to beat average specialists for at least some specific tasks, but uptake of CAD systems and the like has been slowed more by culture than V&V of these approaches.