Thanks for your input on this. I believe that there is still a chance for having centralized AI along with complete privacy guarantees. That’s what I tried to tackle in the other post. Also, the academic community in AI (especially in deep learning) is generally open sourcing most of the newly developed techniques. So that makes the path to AI democratization less steep. Still, there are other issues, like the data availability and the computational resources required. But these will be mitigated eventually.