The brain-computer analogy hampers advancements in artificial intelligence, says Starmind co-founder Pascal Kaufmann.
Starmind co-founder Pascal Kaufmann strongly disagrees with labeling the brain a "computer," with potential for a breakthrough by making it "faster." Kaufmann tells TechRepublic's Dan Patterson that there is, however, a "brain code" that rules governing inside the brain, which can be built artificially, if the principles of intelligence can be extracted. The following is an edited transcript of the interview.
Dan Patterson: If the brain is a small data machine, if consciousness is small data, how do we crack that brain code? Why, then, are we spending so much time, effort and money with big data, and plugging systems you think are counter to the way the brain works?
Pascal Kaufmann: There is a simple reason for this analogy the brain is a computer. A few hundred years ago, people said: in the brain there are pumps and wheels, and levers, because this was the most modern thing that they knew back in the days.
Today it's a computer. People think the brain is a computer, and if you build faster computers, all of a sudden they turn intelligent. You just have to feed more and more data, and then somehow they get intelligent.
The brain-computer analogy is a really bad one. It even hampers progress in artificial intelligence. Therefore, it's so attractive to think the brain is a computer, because it's an input-output machine and you just have to build faster computers ,and then you're there. Unfortunately, in the last 50 years, there wasn't a lot of progress in AI. It seems that this analogy is completely wrong.
Dan Patterson: That we can probably agree with, or at least that maybe of our current metaphors are probably very wrong. But what's right?
Pascal Kaufmann: No one knows exactly what the brain code really is. However, I think that there is a brain code. Meaning there are two, three, four simple rules governing inside the brain. And if you understand them, you can actually extract the principles of intelligence, and build it artificially.
Cracking the brain code is an initiative these days maybe neuroscientists are interested in. I would say about 98% of people in AI believe the brain is a computer, and you just have to build faster computers for a breakthrough. I do not believe in that.
I belong to the second group of people who say we qualitatively do not understand the brain yet. The first camp says we actually have understood it already we just have to build faster computers. I do not believe in that.
- AI and machine learning: Building use cases and creating real-life benefits (ZDNet)
- How deep learning helps match the right patient with the right healthcare (TechRepublic)
- Microsoft advances AI and machine learning through cutting-edge medical research (TechRepublic)
- How one AI company is bringing medical care to millions of rural Chinese residents (TechRepublic)
- AI that improves healthcare efficiency also threatens profits (TechRepublic)
- How to become a machine learning engineer: A cheat sheet (TechRepublic)