He's an AI researcher, tell him to read 2001 by Arthur C. Clake. He doesn't need to read "the classics" he needs to read something proximal to his interests and then figure out which strands he'd like to follow from there.

"The Classics" are at this point literary archeology: they've influenced a lot, but unless you're actually inspired to follow that chain back they're just going to wind up seeming "samey" to things that came after because they were already heavily referenced, extrapolated, deconstructed and their essential ideas endlessly debated.