AI pioneer Marvin Minsky dies at 88

Read the full well-written obituary in Monday's edition of the New York Times

Wired is an American monthly magazine, published in both print and online editions, that reports on how emerging technologies affect culture, the economy and politics. On hearing of the death of Marvin Minsky the editors did a wonderfully appropriate thing: they asked a robot to write his obituary. The software used was Wordsmith, produced by Automated Insights ("Ai") which specializes in analyzing patterns in big data and turns them into readable narratives (in 2014 Ai's output was over one billion stories, more than the combined output of all major media companies combined). Here's what Wordsmith came up with:

Marvin Lee Minsky, 88, passed away January 24, 2016 in Boston, Massachusetts of cerebral hemorrhaging.

Born August 9, 1927 in New York City, New York, to parents Fannie Reiser and Henry Minsky, Marvin Minsky was known for his pioneer contribution to the field of artificial intelligence (AI). After graduating from Phillips Academy, Minsky attended Harvard University, graduating with a BA in Mathematics in 1950. He continued his education at Princeton University, ultimately graduating with a a PhD in Mathematics in 1954.

Some of Minsky’s greatest accomplishments include founding the MIT Computer Science and Artificial Intelligence Laboratory in 1959 and authoring many groundbreaking books in the field of artificial intelligence, including Perceptrons. He won many notable awards in his field of study, including the Turing Award in 1969.

Minsky is survived by his wife Gloria Minsky; three children, Margaret Minsky, Julie Minsky, and Henry Minsky.

Web of Stories, originally known as Science Archive, was set up to record the life stories of scientists and has an online collection of thousands of autobiographical video-stories. There are more than 150 short clips (1-4 minutes each) by Marvin Minsky, two of which are on computer chess:

A short history of chess playing machines

Transcript: Yes I think in 1950 or so, Claude Shannon wrote a… an essay about how to make a machine to play chess. And in retrospect, it’s quite obvious what you have to do. You have to generate possible moves and you have to decide which of two moves is better, but you can’t decide that easily except to see what other moves… subsequent moves result… could result from it. And this was a beautiful paper describing exactly how to make such a machine. There was in fact a similar essay, which hadn’t been widely published, by Alan Turing, the… the scientist in England who also made major discoveries about the foundations of computation.

I don't know what to say. Computers were still too slow at the time that Shannon wrote this paper to make a chess playing machine but which – now we’re talking about the 1950s – but by 1960 the computers were fast enough and we… we and other people wrote computer programs that did more or less what Shannon had described. And we had the first… the first chess playing machine – I think – was made by Allen Newell and Herbert Simon. And the computer was still too small for a whole chessboard, so it played a sort of… miniature kind of chess on a 6x6 board instead of an 8x8 board. And it beat a secretary who had been taught the rules of chess in the hour previous to this match. So, it was the first machine victory, but I don't think anyone took it very seriously.

And then, a couple of years later, John McCarthy invented a strategy which is – for various reasons – is called alpha beta. Which makes the classical chess machine much more efficient and using this trick, by 1960 the simulated chess machines were starting to beat really good players, here and there and... no machine became chess champion of the world until the 1990s, but in the 1960s a… program to play checkers beat the state champion of checkers of Connecticut, a… a large state nearby Massachusetts here. And that was a notable event because no machine had beat a champion player of… at anything before that time.

Did the chess playing machines have an impact?

I think the result of the machines being good at chess and checkers and... eventually, a backgammon machine program developed at Carnegie Mellon Institute by a researcher named Hans Berliner. It beat the world champion at backgammon, for what that’s worth, but I don't think any of these had any serious effect on… on the more general goal of making a machine that had something like human general intelligence: the resourcefulness… ability to solve all sorts of different kinds of problems.

The trouble with these particular games is that you discover a few tricks that… at which machines are better than people because the machine is faster at doing relatively simple things than the person and... so, defeating a person in certain activities... it’s almost like a steam shovel and a person with… with an ice pick. If there’s no real difference except speed, then one isn’t very impressed if a machine... speed or strength. And these chess machines were not particularly clever in any important sense. They couldn’t solve other kinds of problems.