AI is developing at a rapid pace and is becoming very sophisticated. One aspect will include the methods of interaction between AI and humans.
Currently the interaction is an elementary interaction of voice and visual text or images.
Is there current research on more elaborate multisensory interactions?
This is one of the main research areas of my lab which researches intelligent prosthetics which also give sensory feedback such as touch and kinaesthesia (the feeling of a limb moving in space) to the user. We use reinforcement learning to bridge the gap in control and have preliminary work in communicating to the user predictions made by the artificial agent.
Probably these days it's still under the umbrella of "man-machine interaction" in CS, i.e. there is a (sub-) field for interactions between humans and machines in CS, but I am not aware that it has split again to create a sub-sub-field for AI/man interactions.
At risk of seeming out-of-scope, I think it's worth mentioning there may be a case that human vs. AI play of serious games games, such as Chess and Go, represents the "deepest" level of human/AI interaction to date. (Game theory is also important because it underlies all optimization, including, I have no doubt, expanding and optimizing AI-to-human interaction.)
I bring this up because in terms of human-to-computer interaction, I doubt there is a more effective means of engagement than computer games, whether "serious" or otherwise. Thus you might find it useful to look at the concept of Gamification, which naturally lends itself to the type of multisensory input and output you're interested in:
"Gamification can improve an individual's ability to comprehend digital content"
