Honestly I think it can and will be a revolutionary technology once certain enabling breakthroughs are made that allow it to interact more meaningfully with other systems without human intervention, and creating learning data sets that work for each use case with as minimal error chance as possible.

As always though, it's capitalism that's the real problem not the technology itself. Under socialism, we could be automating absolutely everything, saving up our labour power to deal with other problems and correcting the AI's occasional mistakes, but under capitalism any use case will only be implemented because it's profitable, and all the labour power saved will be wasted in the standing army of the unemployed working class.

Also under capitalism making sure that the AI is safe is very costly so the few of us that have to be sacrificed will be worth it for them to rack in the profits.