Thursday, July 18, 2013

Human memory, conversation with computers, and just in time knowledge (and a short demo)


In the end, all our problems in the schools and in education in general, come down to this false belief: “knowledge is important, so we need to tell students lots of stuff.”

Of course, knowledge is important, it is the kind of knowledge that is at issue. It is also the means of acquiring knowledge that is at issue.

Lecturing, reading, memorizing, all that school stuff isn’t really all that natural or all that comfortable for people, and it isn’t how we naturally learn.  

We learn naturally when we try to find something out for ourselves because we are interested in something, or because there is something we are trying to accomplish and we need help. When we want to know how to get somewhere, or how to do something, we ask. Now, typically this has meant asking a person, but more often these days there are ways to ask a computer the same things. And although computers are very bad at conversation, it is nevertheless true that conversation is at the core of how we learn.

We express ideas; others challenges those ideas; we try something; someone makes a suggestion about how we can do it better. We have a problem and we ask for advice. Or we try to make our own thoughts more coherent by trying to express them clearly and dealing with other people’s reactions to whatever we have said.

Most of my professional life has been devoted to studying how human conversation actually works. I have always dreamed that computers would become a part of that conversation and have worked over the years to make that happen. Unfortunately, what passes these days for conversation with a computer is pretty one sided. One reason for that is that computers don’t have that much to say. You might ask, but good luck with finding a reasonable answer on line. The web is full of many interesting things, but finding just the right person who knows the answer to your question and is ready to talk with you about it, is far from reality.

Well, not for long.

Yesterday, my company, Socratic Arts, posted this:


Go there and you will see a short demo of a system that has 1000 video clips of some of the best doctors we have in the U.S. talking about what they know. The clips are short, and the system we use to organize them is based on the book Dynamic Memory (which I wrote in 1981.) Human memory is self organizing. Getting a computer to organize what it knows and re-organize what it knows according to principles of abstract indexing is something we have been working on for years, and it is anything but simple.  

We haven’t been able to show what we have built very easily before, because the clients who sponsor or have sponsored this work usually want their knowledge to be kept private. But this medical project is meant for the general public. It is far from finished. But I thought people should start thinking about what real knowledge looks like: on a computer, in the human mind, and as part of a conversational system that enables knowledge to be delivered just in time (and not ten years before you might need it). 

So have a look.

(This is just an early version of this. A very different one will be available in a month or so. We will keep you posted.)

No comments:

Post a Comment