BRAINPOWER

Everything is interconnected.

“As an ex-programmer I’m still just curious about how the brain functions, how that flow of information really happens,” says Allen in a rare interview, in a conference room overlooking an active ship canal. “The thing you realize when you get into studying neuroscience, even a little bit, is that everything is connected to everything else. So it’s as if the brain is trying to use everything at its disposal–what it is seeing, what it is hearing, what is the temperature, past experience. It’s using all of this to try to compute what the animal should do next, whether that animal is a mouse or human being.”

My question is: Why aren't more people investing in this???

Now Allen, the 20th-richest man in America, with an estimated net worth of $15 billion, has committed another $300 million for projects that will make his institute more than just a maker of tools for other scientists, hiring several of the top minds in neuroscience to spearhead them. One effort will try to understand the mouse visual cortex as a way to understand how nerve cells work in brains in general. Other projects aim to isolate all the kinds of cells in the brain and use stem cells to learn how they develop. Scientists think there may be 1,000 of these basic building blocks, but they don’t even know that. “In software,” Allen says, “we call it reverse engineering.”

To answer your question: Because this is most likely basic research, where the primary benefits will flow to people using the information (assuming they find anything at all :-). Investors prefer things where they see a way to capture most of what they find. :-(

Also, my bet is that we'll actually find out more by pursuing Artificial Emotion rather than Artificial Intelligence of Artificial Neurons. But even Paul Allen isn't crazy enough to invest in that. :-P

Thanks Ernie.

Artificial emotion sounds like something that could make Siri addictive.

I want to hear Siri throw a shit fit and complain that you're always asking her for stuff and never think about what you can do for her. ;)

Not exactly, but close. My view is that we should build computer systems with "artificial emotions" regarding its Health (subsystems), Happiness (satisfying the user), and Holiness (impact on other systems). These could be modeled crudely as various indicators that influence behavior. For example, a computer that hogged a lot of bandwidth to download a movie would "feel bad" and let the other systems on the network have priority.

Interestingly, to do this well requires some sort of predictive anticipation.

Excitement: "I know my owner is taking me on trip tomorrow, so I should ensure my battery is fully charged and my local cache is up to date."

Depression: "I am being ignored, so I may as well power down into battery-maintenance mode to extend my life."

One can even imagine systems using downtime (dreaming?) to rehearse usage patterns to detect common pathways to optimize...

All the pieces to do this are there, and it would eventually allow our systems to become far more efficient and responsive. But alas, computer scientists barely acknowledge the validity of their *own* emotions, much less their utility for our slave machines....

Ernie, that sounds awesome. Siri crossed with Tamagotchi!

Kind of like Marvin in The Hitchhiker's Guide to the Galaxy...