I question whether it ever really will. Despite significant financial and time investments, much of it still generally remains useful to specific tasks. If you need 50 robots to do 50 different things because diversifying their usefulness remains a pipe dream, it's a tough sell. It's also a tough sell whether the majority of people actually want and/or see a benefit to their life on the consumer side in addition to a growing consensus that, in the case of GenAI, what it outputs still kind of needs verification and is about as trustworthy of a source as Wikipedia. The latter statement, in my opinion, will be a hard one to break since it has started off so poorly.