For computers, beating the world's best chess player is much easier than knowing when to smile. Although we take such subtleties for granted, we know immediately when someone gets it wrong. Such errors affect our perceptions of "intelligence" -- and will, I think, be a fundamental issue in determining whether we see computer-modeled brains as conscious or human-like.
In the quest to model the human brain in computers, the place of artificial intelligence is uncertain. On one hand, it's what comes to mind these days when you talk about computers that "think" and "learn" as people do. But as AI experts are tired of pointing out, not all AI techniques are accessible to human brains.
I've been busy working on a 45-second "teaser trailer". Here's a 10-second sneak peek of just the tagline.
I got the idea from "Almost a Brain" after learning that the world's top supercomputers are nearing the level of raw power in our brains — about a billion billion (1,000,000,000,000,000,000) operations per second (an "exaflop"). I've since come to recognize that there's a lot more to brain modeling than power, but supercomputing still has a special place in my heart.
And why they won't be featured in "Almost a Brain"
Last Friday I announced my upcoming documentary about computers that model the human brain, "Almost a Brain". And I've started talking about it to everyone I can. I practice my pitch on them while watching their faces for signs of interest, skepticism, and outrage. This is market research; it'll affect what the documentary covers, and how.