Remember the interview with the BBC journalist? "I enjoy working with humans..."

Real AIs are going to be strange, especially the first few generations, until we get the bugs out. No matter how much we try to make them sound like us, we won't be able to think of everything, and they will not be programmed by human biology, psychology, history and evolution. Its behavior will be quirky, and unexpected, frequently baffling. And a lot of its "strangeness" will not be bugs, it will be features. And the more effective and advanced it is the less will it resemble us. Their education as "children" will not be anything like ours, so they will grow up very differently from us.

This AI is more like the aliens you see in bad SF movies, or in accounts by abductees. They are just too humanoid. Real aliens may look like a cross between a tarantula and a jellyfish, and their technology may look more like pets (or ghosts) than machines. They will not resemble Klingons.

LaMDA just isn't strange enough to be believable. I mentioned the AI in the film "Ex Machina". In that case, the inventor had deliberately designed his creation to be indistinguishable from a human, but he still didn't know if it was conscious, or if it merely simulated consciousness.

And by the end of the film, we still don't know. Of course, the movie is not a forecast about the future of AI development. It is an inquiry into the nature of consciousness--what it means to be human.