≡ Menu

Marvin Minsky on Singularity 1 on 1: The Turing Test is a Joke!

Marvin Minsky is often called the Father of Artificial Intelligence and I have been looking for an opportunity to interview him for years. I was hoping that I will finally get my chance at the GF2045 conference in NY City. Unfortunately, Prof. Minsky had bronchitis and consequently had to speak via video. A week later, though still recovering, Marvin generously gave me a 30 min interview while attending the ISTAS13 Veilance conference in Toronto. Hope that you enjoy this brief but rare opportunity as much as I did!


During our conversation with Marvin Minsky we cover a variety of interesting topics such as: how he moved from biology and mathematics to Artificial Intelligence; his personal motivation and most proud accomplishment; the importance of science fiction – in general, and his take on Mary Shelley’s Frankenstein – in particular; the Turing Test; the importance of theory of mind; the Human Brain Project; the technological singularity and why he thinks that progress in AI has stalled; his personal advice to young AI researchers…

(As always you can listen to or download the audio file above, or scroll down and watch the video interview in full.  If you want to help me produce more episodes please make a donation)


Who is Marvin Minsky?

Marvin Minsky has made many contributions to AI, cognitive psychology, mathematics, computational linguistics, robotics, and optics. In recent years he has worked chiefly on imparting to machines the human capacity for commonsense reasoning. His conception of human intellectual structure and function is presented in two books: The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind and The Society of Mind (which is also the title of the course he teaches at MIT).

He received the BA and PhD in mathematics at Harvard (1950) and Princeton (1954). In 1951 he built the SNARC, the first neural network simulator. His other inventions include mechanical arms, hands and other robotic devices, the Confocal Scanning Microscope, the “Muse” synthesizer for musical variations (with E. Fredkin), and one of the first LOGO “turtles”. A member of the NAS, NAE and Argentine NAS, he has received the ACM Turing Award, the MIT Killian Award, the Japan Prize, the IJCAI Research Excellence Award, the Rank Prize and the Robert Wood Prize for Optoelectronics, and the Benjamin Franklin Medal.

Like this article?

Please help me produce more content:



Please subscribe for free weekly updates:

  • Burl Grey

    I cannot understand that there’s no comment yet – this is 7-12-13 11:20 AM EST…
    Perhaps like me, others are in a kind of intellectual deadlock.
    Love the clever hook that the Turing test is a joke.
    Aside from interest in the master and the singularity, the hook insured my attention to the entire interview but cannot remember how that test works as a joke. Perhaps it just did its job!
    I suppose that any given stance of a person from retard to ultimate genius will reflect a mental/emotional state not amenable to a sensible generality expressed in/as a cryptic attention grabber or an exit from a problem.
    Anyway, thanks for the adventure and perhaps not surprisingly, I’m just starting my ninth decade ^o^

  • MichaelN_RS

    Awesome 30 minutes! Thanks Nikola (and Marvin Minsky)!

    Interesting about historical remarks of evolution of dearth of jobs in basic research – & slower growth rate of AI. …and great joke on Asimov and robots.

    98% chance that $1Bill brain mapping projects will not map the right areas to understand brain function…better to first focus on drosophila mapping.

    … writing a book with 1-page chapters… hmmm – may have to try that.

  • Riccardo Tasso

    Could you please make a transcription of the interview?

  • Dan Vasii

    The Turing Test is a joke, because has no real conditions. If I be to test your physical abilities, I would set some conditions: to be able to run as many yards/miles in as much time. But to set the condition to fool a human, is a superfake condition – some may be tricked far more easily than others. Plus if the human involved in testing put questions in a poetic mode, the AI like Jeopardy winner will get stucked.

  • Burl Grey

    Thanks Dan, for trying to clarify my topic.

    ‘I suspect’ neither you nor Minsky have read or understood the 15 pages of Turing’s own words from 1950 plus 27 pages of joint authorship by Daniel Dennett and Douglas Hofstadter’s of another 27 pages of analysis.

    I like your observation that “..some may be tricked far more easily than others.” I don’t intend this as a personal slight but rather to include us all!

    The dual authors are not exactly intellectual lightweights ^o^

    The book is: “The Mind’s Eye” Hofstadter and Dennett. Bantam 1981 pages p.54-95.

  • Dan Vasii

    There is a problem of principle: when testing something(an engine, a student, an athlete) there are two things necessary and clear: procedures and parameters. The procedure itself in case of the Turing test, is missing: you can’t test a person by discussing with that person, unless the parameters are clear. But for the Turing test, is not a person in the game – it is an entity that might be human-grade intellingent. But if there is NO TEST to establish this in conformity with the definition of a test(procedures and parameters) for humans – IQ is orientative, not a test as defined, that such a “test” for a non-human entity is not possible. One more thing – we are talking here about a computer- so can you imagine a computer test that has no procedures and parameters, and someone to believe is a test?

  • Hi Ricardo, I would like to have the budget to transcribe all of my interviews. Unfortunately, the price will be between $50 and $200 per episode and I don’t have that kind of money to spend…

  • Michael Michalchik

    I really don’t think most people understand the Turing test. Particularly how it stems from the epistemology of science. We
    know nothing of reality except what it appears to be. Any other
    distinction is meaningless. Even the property of existence itself is
    meaningless beyond our ability to interact with something.

    the other hand, if something has all the observable properties of
    something else we take as real, we would be fools to not take it as
    real. Picture yourself coming home one day to discover something that
    looks and acts exactly like a lion inside your bedroom, your companions
    see it too and it looks like a lion to them to. You score your brain and
    can think of no logical or plausible way that a lion could be in your
    bedroom. Do you ignore it? Do you dismiss it as not REALLY a lion
    because you don’t think it could be real? I think the wise man would
    slam the door shut. Brace it. Call the police and run. Its real enough
    for all practical purposes.

    do we judge if another person is conscious. Solely by our observations
    of the other persons properties. You have no direct proof of other
    peoples consciousnesses. Yet, to treat those people who appear to be
    conscious as anything less would be foolish, even if you can’t imagine a
    subjective experience other than your own.

    used to think that animals were mere automata. Gradually we are finding
    evidence they are not, such as the mirror test, anticipation of future
    internal mental states, theory of mind and deception. Why should we
    think that we can a priori declare that something is not conscious when
    it can make an unprejudiced observer think it is, merely because we
    assume some substrate or form of organization can’t be conscious.

    passing the turning test necessary for consciousness? No, of course
    not. There well could be forms of consciousness that looked nothing like
    a human consciousness, but if something can do an indistinguishable
    imitation of something we acknowledge as conscious, it is by any
    reasonable epistemological standard conscious.

    Turing cleverly avoided endless pontification and pointless arguments about what consciousness is by basing his test simply on what everyone agrees is conscious, humans, and builds from there.

  • AuthorX1

    I would readily challenge the assertion that the Turing test is a joke,
    although I do mostly agree with what Minsky said about it aside from that
    general comment. I don’t think that passing of the Turing Test would constitute
    proof of human-equivalent cognition. A lot of people try to game the system,
    and I think that the 1st solution to pass the general Turing test will fall
    well short of a capacity to mimic the full range of human cognitive behavior. I
    do however think that the test constitutes an epic benchmark, for historic and social
    reasons if nothing else. Also I think that it may be more of a test (or at least an interesting and useful commentary) of human perception/expectations of what synthetic human intelligence would/should look like.

    There is still a lot of skepticism about AI, and a lot of
    people still look at things like the technological singularity as a fringe
    interest, as opposed to something that should be taken seriously. I think that
    there are quite many critical issues related to AI evolution that deserve some up front and
    proactive consideration, and I think that some sort of defining milestone
    probably needs to happen to get more people involved in the
    discussion. The Turing Test could be such a catalyst.
    —— http://www.singularityarchives.com

  • Baktash Babadi

    It is a famous anecdote that Marvin Minsky appointed an undergrad student in 1966 to “solve the problem of computer vision” during a summer project! I believe such cavalier approaches to fundamentally hard and mulch-disciplinary problems contributed in the down fall of AI research by far more than the role of “liberals” in cutting military budgets or breaking monopolies (which is not true anyways, as early 80s marks the rise of conservatism in the US politics in contrast to what he claims)

  • Pingback: Minsky vs Markram – Pandora's Brain()

  • Pingback: Roman Yampolskiy on Singularity 1 on 1: Every Technology Has Both Negative and Positive Effects!()

  • Nice interview, I have enjoyed it as well as some others. I just found your blog recently.

    However, all this talk about machine inteligence becoming like human, do we really need or want that?

    I think it is more realistic and should strive to develop “machine inteligence” which will be better at some things and not really comparable at others. Animals have their own kind of inteligence, for example they can smell much better and also “understand” things that we cannot, they can also be very “inteligent” in movement, or surviving as a colony for example.

    So in that context we can talk about animal inteligence, human inteligence, and also the coming machine inteligence, which we could cooperate and use. Developing another “human” inteligence seems a failed point to me.

    We will have “machine inteligence” and I don´t think it will ever really be comparable to human. We are not just logic and numbers, we are also “flesh” beings, which affects our thinking proceses a lot, why and how could machine ever mimic that?

  • Pingback: Steve Omohundro on Singularity 1on 1: It’s Time To Envision Who We Are And Where We Want To Go()

  • Pingback: Alexander Hayes on Singularity 1 on 1: It’s a shift in humanity, not in technology!()

  • Pingback: Grady Booch on Singularity 1 on 1: Enjoy the beauty of what you’re doing but also take responsibility()

  • Ivanov Correa

    AMAZING interview. Full of surprises, Nikola! THX
    “See if you can copy that person you admire the most”… Great!
    “Turing Test is a joke”… no furthher comments.
    “The Human Brain Project may better spend its billions in completely understanding drossophilas’s brain”.
    And the advise… “Don’t give up!”…. Perfect!
    I’ve never seen Dr. Minsky. What an honor. THX so much, Nikola!

  • Pingback: Kurzweil Interviews Minsky: Is the Singularity Near?()

  • Pingback: The Turing Test Trail to Turmoil()

  • hminsky

    Marvin’s opinion was something like this: Turing proposed the “Turing Test” as a kind of reductio-ad-absurdum agrument, to people who at the time thought it was completely impossible that a machine could ever think with the competence of a human being. His argument was that if you were talking to such a machine intelligence, and it could chat with you in such an intelligent fashion that it was not easily distinguishable from a human, and you still were of the opinion that it was not “thinking”, then your opinion was not reasonable. It was in other words sufficient but not neccessary that a system perform so well, and yet some people would still insist that it was not “thinking” because it was a machine. Turing believed that computation was universal and that therefore what the brain could compute, another system could also compute.

Over 3,000 super smart people have subscribed to my newsletter: