It's difficult to pin down what exactly qualifies as AI - are the deep analytics that Amazon uses for marketing purposes, for example, AI? Are driverless cars "intelligent", or are they just programmed with a huge number of scenarios and act programatically?
As we connect more and more things to the internet, it's becoming a giant robot, with the ability to gather information through sensors, modify things in the real world through APIs to physical devices, and so on. But these things need to be controlled, what is the management structure.
There are interesting identity problems to solve around AI, and will get more complex. Things like giving consent to AI to make decisions on your behalf (e.g., trolley problem with autonomous cars)
When looking at authentication, AI could be helpful with determining the authenticity of a biometric response to increase the amount of data that can be considered.