NEXT STORY
Developing ideas of intelligence in the 1960s
RELATED STORIES
NEXT STORY
Developing ideas of intelligence in the 1960s
RELATED STORIES
Views | Duration | ||
---|---|---|---|
71. Freeman Dyson proves what I couldn't | 4531 | 03:17 | |
72. Solving Emil Post's problem | 3494 | 03:00 | |
73. Solving Post's unsolvable problem leads to the 'Minsky Machines' | 1 | 3322 | 03:49 |
74. Andrew Gleason's eight year plan of attack | 3138 | 03:40 | |
75. 'You can pull something with a string, but not push it' | 2983 | 02:02 | |
76. Will machines ever understand Aesop's fables? | 2926 | 00:51 | |
77. Analogy is the difference between human and computer thinking | 2 | 3067 | 05:07 |
78. Developing ideas of intelligence in the 1960s | 2577 | 02:01 | |
79. Unexpected problems with language machines | 2574 | 00:50 | |
80. The history of the laws of physics | 2774 | 02:44 |
One question is why, if we were making such rapid progress in the early 1960s, solving college level problems in... in scientific fields, why can’t we do... 60 years later, 50 years later anyway, why don’t we have programs that could answer the kinds of simple questions that any two or three year old could do and I've a number of ideas about that. In fact I’ve written two books about what I think one would need to achieve a certain level of common sense thinking ability and I have a bunch of theories about why we don’t have such systems and the most interesting question is... is simply, if you observe that – so far as I know – there are only about 20 or 30 people who are being supported to spend full time making theories of normal, smart human reasoning.
There are tens of thousands of people who are working on particular ideas about that, rather than general ideas. For example, how do you make a system that evolves by trial and error to simulate certain aspects of evolution or how do you make a machine that tries to calculate the probability of certain kinds of events, given a large amount of data about related events. But there are almost no groups or individuals working on how do you make a program that makes really productive analogies so that if it sees one thing working it can say, I know how to apply something like that to a somewhat different problem. It seems to me that we have tens of thousands of highly educated people working on trying to apply logic to thinking and it’s just the wrong thing to do. I don’t think people use anything very much like logic. What they do is a lot of analogy. This situation resembles this one in some ways but not others, which of these ways is important for solving the problem I have, let’s reformulate and find a new description of this that emphasizes the features that might be relevant to the new problem I’m trying to solve.
This is the sort of thing you... you probably do something rather complicated like that every two or five seconds. When you walk through a room, you recognize certain kinds of things as... as harmless, other things as obstacles that you’ll have to get around, yet other things as being particularly fragile or expensive, you don’t want to damage them and... a great many different kinds of awarenesses, only a few of which actually come to the level where your articulately expressively thinking about them in relation to some other problem that you’re trying to solve. And as far as I know, there are only about 10 or 20 or maybe 30 people in the world who have the leisure or opportunity to try to make theories about common sense ordinary thinking.
We have tens of thousands of paid... highly paid to try to make predictions about the stock market or about the not stock market. How many groceries should I order for my store? Is there a business system I can buy from IBM or Microsoft or somewhere that will allow me to increase my profit by the 1%, which is the difference between gradually losing things and gradually gaining them? Tremendous energy on just barely significant applications that allow us to get by rather than fail, very little work on saying, how can I... how can I understand intelligence so that I could save children four of the 12 years of elementary school? Why... why can’t they go to college after eighth grade? We’re spending a large fraction of everybody’s life on things that haven’t changed for 100 years and nobody’s being paid to work on those sorts of questions.
Marvin Minsky (1927-2016) was one of the pioneers of the field of Artificial Intelligence, founding the MIT AI lab in 1970. He also made many contributions to the fields of mathematics, cognitive psychology, robotics, optics and computational linguistics. Since the 1950s, he had been attempting to define and explain human cognition, the ideas of which can be found in his two books, The Emotion Machine and The Society of Mind. His many inventions include the first confocal scanning microscope, the first neural network simulator (SNARC) and the first LOGO 'turtle'.
Title: Analogy is the difference between human and computer thinking
Listeners: Christopher Sykes
Christopher Sykes is a London-based television producer and director who has made a number of documentary films for BBC TV, Channel 4 and PBS.
Tags: 1960s, IBM, Microsoft
Duration: 5 minutes, 8 seconds
Date story recorded: 29-31 Jan 2011
Date story went live: 12 May 2011