Nanotech for Info Management

A WSJ interview with IBM’s Gerd Binnig who won a Nobel Prize for his work on nantech. He is now working on “helping people better tap into the vast quantities of information stored in databases, hard drives and the Internet” by building nano-machines to replicate the human brain. He explains the problem…

The problem is that [there’s] simply too much [information] for us. We need some help from machines that predigest this information. This means these machines to a certain extent have to understand this information.

If you work on a certain problem in science and you stumble over something new, you have to read an awful lot of scientific literature to see what has been written about similar or related phenomena. I would say 90% of the literature you wind up looking through isn’t of interest to you, but you simply have to read it all, as you do not want to miss something important.

Instead, you could have an intelligent assistant, which could be a machine, that reads all the literature for you and gives you just the information you need, like, “Here in this paper they talk about this and that,” and brings up a connection between two different subjects. Then you can read this predigested information and pick up the right papers and look only at them and not all the vast amount of literature.

…and points to a possible solution:

The present way of processing data is still very useful. We, however, will learn more and more from the brain. I envision for the processor of the future a complex hybrid of conventional sections and brainlike ones. Let’s take the example from before of understanding text. You show a computer a word. You can say this word is known in the computer’s knowledge base … but it has many, many meanings.

In the neighborhood of this original word, you found other words, and you also don’t know the meaning. You have to link these words in a logical way sometimes, and sometimes only in a very loose, associative way. So they have to influence mutually their meanings. This could be done in a very parallel fashion, where you stimulate all of these words and at the same time they stimulate each other in defining their meaning.

For this you need lots and lots of connections so they influence each other. Then you have not only the words that influence each other. It might be that the previous sentence has to influence it. So you build up a meaning on a higher level, a more abstract meaning that can be digested somewhere else, which also influences then this particular sentence you’re analyzing.

Published by

Rajesh Jain

An Entrepreneur based in Mumbai, India.