Artificial Intelligence Mind

Polanyi’s Paradox: Why machines can’t think as we do

Spread the love
Michael Polanyi (1891-1976)

From Mind Matters today:

Recently, we looked at Moravec’s Paradox, the fact that it is hard to teach machines to do things that are easy for most humans (walking, for example) but comparatively easy to teach them things that are challenging for most humans (chess comes to mind).

Another paradox worth noting is Polanyi’s Paradox, named in honor of philosopher Michael Polanyi (1891-1976), who developed the concept of “tacit knowledge” …

Here’s [Polanyi’s] Paradox, as formulated by law professor John Danaher, who studies emerging technologies, at his blog Philosophical Disquisitions:

We can know more than we can tell, i.e. many of the tasks we perform rely on tacit, intuitive knowledge that is difficult to codify and automate.

We have all encountered that problem. It’s common in healthcare and personal counseling. Some knowledge simply cannot be conveyed—or understood or accepted—in a propositional form. For example, a nurse counselor may see clearly that her elderly post-operative patient would thrive better in a retirement home than in his rundown private home with several staircases.

The analysis, as such, is straightforward. But that is not the challenge the nurse faces… More.

Claims about machines putting people out of work, as the article goes on to discuss, must be evaluated in the light of this problem.

See also: Why can’t machines learn simple tasks?: They can learn to play chess more easily than to walk. If specifically human intelligence is related to consciousness, the robotics engineers might best leave consciousness out of their goals for their products and focus on more tangible ones. (Moravec’s Paradox)

4 Replies to “Polanyi’s Paradox: Why machines can’t think as we do

  1. 1
    PaoloV says:

    The talk about “conscious” AI is a complete nonsense.
    You may add mustard and ketchup, but it remains nonsense.

    What affects what?
    What causes what?
    How?

    What is agape love?

    Check this out:

    https://www.parents.com/baby/all-about-babies/snuggling-your-baby-may-affect-their-dna-says-a-new-study/

    In the beginning was the Word, not the printer, not the paper, not the ink, not the audio devices, not the video technology, not matter, but Logos, pure Logos.

  2. 2
    daveS says:

    We have all encountered that problem. It’s common in healthcare and personal counseling. Some knowledge simply cannot be conveyed—or understood or accepted—in a propositional form. For example, a nurse counselor may see clearly that her elderly post-operative patient would thrive better in a retirement home than in his rundown private home with several staircases.

    This is all true, but my understanding is that when using neural networks, for example, this sort of tacit or intuitive information does not need to be expressed in propositional form in order for the AI to learn it.

    You “just” have to gather the data and train your model on it.

    I put quotes around the “just” because it’s still a massive amount of work, both for human data scientists and the machines, and success is not guaranteed.

    Now I’m not expecting my physician to be replaced anytime soon, but I would not be surprised to see AIs soon achieving better results than human practitioners in some carefully chosen tasks, for example analyzing imaging tests such as mammograms, etc. [Edit: Looking further, it appears they already can.]

  3. 3
    daveS says:

    PS: The example of the elderly patient living in a multi-storey home might not be the best for illustrating the tacit, intuitive knowledge that a human could possess over a machine. Medical professionals are extremely aware of the danger of patient falls, and patients are routinely asked about their home environment when undergoing major surgery. Here’s a questionaire given to patients after hip/knee replacement, as an example.

    I don’t have a better example at hand, but perhaps something like a Columbo-esque flash of insight, where he solves a case by picking out some apparently minor detail which connects the suspect to the crime.

  4. 4
    vmahuna says:

    A million years, back when dinosaurs still walked the Earth and I was a new hire in a government office, EVERY manager had a clerk/typist. You could fill out a Memorandum form in handwriting, but of course all of the important stuff had to be typed on a real live typewriter. By the end of the 1980s, that paradigm had changed, and every competent officer worker (e.g., Analyst) did their own typing on a PC. Practically all of the clerk/typist positions disappeared, although the more senior Managers still had a girl who sat at the “secretary’s” desk outside his office to indicate his importance. She had a PC connected to the network printers.

    This was a HUGE change for office workers because it essentially eliminated the non-college educated staff en bloc. All of the Analysts were already picked from folks with college degrees, and so the elimination of jobs for clerk/typists and file clerks (the “file” is on the network, not in a steel cabinet…) offered no equivalent jobs. As near as I can tell the new equivalent is “miscellaneous paper-shuffler”, a position that includes “making copies” of stuff that everybody invited to the meeting already has on email.

    There was a stunt by one of the billionaires (Rupert Murdock?) some decades ago with the Times of London. It was traditional that the Type Setters Union would go on strike every time their contract ran out. And so the Union was itching for a BIG raise, knowing that without preparation of the physical plates to literally PRINT the newspaper, The Times would lose huge piles of money every single day they had not agreed to a new contract with BIG raises.

    But they were DEAD wrong. The billionaire had secretly assembled a COMPLETE electronic printing plant unknown to anyone in the Union. So the day after the Union went on strike, The Times printed their first “digits to ink” newspaper. The blue collar unionists were STUNNED.

    The Union quickly backtracked and said they would agree to an extension of the old contract. Management said, “Well, no, you didn’t want to negotiate before, and WE don’t want to negotiate NOW.”

    The printers union simply evaporated. And other print companies quickly dropped their Typesetters and went to computers. The typesetter jobs went the way of makers of buggy whips (which is now probably too archaic to even convey any meaning).

    On the other hand, there have been attempts to produce software that can take a string of facts generally related to a topic and produce the equivalent of a reporter’s column. The attempts have generally failed. The CREATIVE part of arranging facts to clearly describe the situation and then delineate the problem and its possible solution are still beyond the scope of Machine Intelligence.

Leave a Reply