Marvin Minsky, Pioneer In Artificial Intelligence, Dies at 88 (nytimes.com)

An anonymous reader sends word that Marvin Lee Minsky, co-founder of the Massachusetts Institute of Technology's AI laboratory, has died. The Times reports: "Marvin Minsky, who combined a scientist’s thirst for knowledge with a philosopher’s quest for truth as a pioneering explorer of artificial intelligence, work that helped inspire the creation of the personal computer and the Internet, died on Sunday night in Boston. He was 88. Well before the advent of the microprocessor and the supercomputer, Professor Minsky, a revered computer science educator at M.I.T., laid the foundation for the field of artificial intelligence by demonstrating the possibilities of imparting common-sense reasoning to computers."

Comments

From his Wikipedia article:

[Anonymous Coward]

"Isaac Asimov described Minsky as one of only two people he would admit were more intelligent than he was."

Wow.

My favorite quote

[tgv]

Heard it from a teacher who had heard it from Minsky, but it's probably not literal anymore, after all those years: consciousness is just a feedback loop.

He was truly one of the greats.

Re:My favorite quote

[John Allsup]

The only issue with the quote is 'just'. Any recursive process is a feedback loop, feeding the state at stage n as input to produce the state at stage n+1. Likewise, any dynamical system is 'just' a feedback loop, the state at time t being input to a set of rules that determine the state at time t+dt (before you do the limiting stuff, though you can use nonstandard analysis to avoid issues with dt being infinitely small).

Something as complex as consciousness would surely be impossible in the absence of feedback loops.

Re: My favorite quote

[khallow]

If you think of the mind/brain as a mechanism, you're already wrong.

Unless, of course, you're not wrong. That's always the rhetorical problem with just asserting stuff. It can be false.

My favorite Minsky story

[Anonymous Coward]

In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6.
"What are you doing?" asked Minsky.
"I am training a randomly wired neural net to play Tic-tac-toe," Sussman replied.
"Why is the net wired randomly?" asked Minsky.
"I do not want it to have any preconceptions of how to play," Sussman said.
Minsky then shut his eyes.
"Why do you close your eyes?" Sussman asked his teacher.
"So that the room will be empty."
At that moment, Sussman was enlightened.

Re: My favorite Minsky story

[maitas]

I read "society of mind" when it first got translated to Spanish. Lovely, lovely book. Each page was a concept by itself. You could literaly read one page a day an have a complete knwoledge exerience each time.

What always makes me feel uneasy of Minsky was the story that he attacked perceptron showing that they could not solve NOR sending neural nets into a freeze for sevwral years drying founding to them, while at that time it was already known that multilayer nets could solve NOR.

Nowadays deep learning is all the rage...

Re: My favorite Minsky story

[maitas]

Sorry about that. It was written on a cell phone and I m not native english speaker.

    Nevertheless here goes the story ( http://www.andreykurenkov.com/... )
Marvin Minsky, founder of the MIT AI Lab, and Seymour Papert, director of the lab at the time, were some of the researchers who were skeptical of the hype and in 1969 published their skepticism in the form of rigorous analysis on of the limitations of Perceptrons in a seminal book aptly named Perceptrons. Interestingly, Minksy may have actually been the first researcher to implement a hardware neural net with 1951’s SNARC (Stochastic Neural Analog Reinforcement Calculator) , which preceded Rosenblatt’s work by many years. But the lack of any trace of his work on this system and the critical nature of the analysis in Perceptrons suggests that he concluded this approach to AI was a dead end. The most widely discussed element of this analysis is the elucidation of the limits of a Perceptron - they could not, for instance, learn the simple boolean function XOR because it is not linearly separable. Though the history here is vague, this publication is widely believed to have helped usher in the first of the AI Winters - a period following a massive wave of hype for AI characterized by disillusionment that causes a freeze to funding and publications.

The point here is that Minsky already knew that a multilayer Neural Network could solve XOR, but he nevertheless wrote a paper showing that a single perceptron could not solve XOR, in order to get funding away from perceptrons and towards his own line of investigation.

All that said, I highly recommend his "society of mind" book. The idea to have each single page to explain a simple concept in a complete manner was mind blowing to me at that time. I still havent come across another book written that way.

Re: My favorite Minsky story

[Chris Mattern]

Minsky wasn't being disingenuous or ignorant. He was well aware that multilayer nets could solve XOR (not NOR, which wasn't a problem either way). His objection was that he thought multilayer nets were not trainable, which was indeed the general opinion at the time.

Now, as you point out, we have deep learning algorithms, but those didn't exist then.

1973

[mbone]

I first saw Minsky give a presentation in 1973, at MIT. It was full of confident assertions that, as soon as we had sufficient CPU power, say by 1980 or so, we would have true AI. It was just around the corner and we would have to get used to its implications, etc.; all it would take would be a few megaflops and more RAM, and that was all improving rapidly.

This was not the last confident presentation I have heard from an AI researcher. It all gives me a certain skepticism about confident AI predictions.

If you haven't read Society of Mind by Minsky

[turp182]

I would recommend it.

http://www.amazon.com/The-Soci...

Very interesting insights into ideas about how consciousness interprets the reality around us (and how the mind ties it all together into something meaningful).

Re:My favorite quote from him:

[justthinkit]

We are least aware of what our minds do best.
- Marvin Minsky

My own Marvin Minsky story on neural networks

[Paul Fernhout]

Posted here first this morning (couple of types fixed): https://ma.tt/2016/01/minsky/#...

Wow, sad to hear the news. Marvin Minsky and I were academic peers of a sort -- he was one of George A. Miller's first students, and I was one of George's last students. :-) George told my parents something like I was the student who most reminded him of Marvin Minsky, except whereas he spent George's Air Force money, I spent my father's money. :-) Which was not quite true (I paid for a chunk of Princeton with the proceeds of a video game I wrote and with some loans) but it sounded funny. :-) My dad was actually mostly a blue collar worker, and my mother only later in life worked for county social services, so George may also not have realized my family was not that well off financially.

I met Marvin Minsky once in his MIT office in 1985 as I was graduating from Princeton. I likely gave him a copy of my thesis -- "Why Intelligence: Object, Evolution, Stability, and Model". I also wrote to him once in the 1990s about getting computer time for space habitat simulations (he was responsive in a positive way, but then I met my wife and so just let stuff like that drop). And I saw him in passing about fifteen years ago when he gave a talk at IBM Research while I was a contractor there (he spoke about multiple simultaneous mental representations, and picking from the best one). A nephew of his even lived down the hall from me my senior year at Princeton in 1903 hall, too, but I never talked with him about his uncle. But we never really connected any of those times, sadly.

One of the biggest mistake I've made in my life careerwise (or so it seemed at the time) was when visiting Marvin Minsky in his office to talk to him about the triplestore and semantic network ideas in my thesis (stuff that indirectly helped inspire WordNet which George started as I graduated). I casually mentioned in passing to Marvin Minsky very early on in our meeting something about neural networks (MIT had a spinoff then of the Connection Machine), and I guess that may have put him in one of those mental states where some of the 400 different little computers activate. :-) I had not known then that he had essentially written a book (Perceptrons) to discredit neural networks (by only considering a limited version of them) to preserve funding for more formal semantic networks he worked on. He warned me sternly about how many careers had been destroyed by exploring neural networks. Another of George's students had found a copy of Marvin's original SNARC paper (what Marvin spent George's Air Force grant money on), and I can wish I had thought to take a copy to Marvin, as that might have set a different tone for our meeting, as it turned out Marvin had lost his original and wanted to reference it in his book "the Society of Mind" he was working on then.

So, instead of MIT, I spent a year hanging out in Hans Moravec's and also Red Whittaker's robot labs, and that was interesting in its own way. That experience also set me to thinking about the implications of most of the CMU robotics work being funded by the US military, which ultimately lead to my key insight about the irony of using robots to fight about material scarcity they could otherwise alleviate.

I sent Marvin Minsky an email in 2010, with a subject of "Vitamin D, computing, and abundance", warning about the health risks of vitamin D deficiency for heavy computer users. I also thanked him for his interactions with James P. Hogan, an author whose writings have been very inspiring to me (like Two Faces of Tomorrow and Voyage From Yesteryear), as James acknowledges Marvin in the first as a major source of ideas and inspirations, so some big ideas went from Marvin to James to me at least in that sense. :-) I also thanked him for being such an inspiration in years gone by. I had been reading through all the comments at a Wired article on "D