6 ms·

Autopilot was invented in 1914. It is clearly a form of artificial intelligence. That AI doesn't require computers should be a conceptual shock-- one that can h

Autopilot was invented in 1914. It is clearly a form of artificial intelligence. That AI doesn't require computers should be a conceptual shock-- one that can help us better understand what AI really is.

throw0101a 5y ago
I'm not sure a feedback loop is "clearly" AI: * https://www.electronics-tutorials.ws/systems/negative-feedback.html https://www.electronics-tutorials.ws/systems/negative-feedba... * https://en.wikipedia.org/wiki/Feedback https://en.wikipedia.org/wiki/Feedback There are no heuristics involved: * https://en.wikipedia.org/wiki/Heuristic_evaluation https://en.wikipedia.org/wiki/Heuristic_evaluation * https://en.wikipedia.org/wiki/Heuristic_(computer_science) https://en.wikipedia.org/wiki/Heuristic_(computer_science)
- bencoder 5y ago
  Seems like an example of the AI effect: https://en.wikipedia.org/wiki/AI_effect https://en.wikipedia.org/wiki/AI_effect although I think when autopilot was invented, the concept of "AI" didn't exist, so maybe not a perfect example
- dr_dshiv 5y ago
  Autopilot is clearly AI. If I didn’t mention 1914, you wouldn’t doubt it. A feedback loop — namely between a system and its measure of its own performance— is central to the idea of AI. At least according to Peter Norvig, the director of research at Google, who defines intelligence as ‘the ability to select an action that is expected to maximize a performance measure’ (Russell & Norvig, 2016, p. 37).
  - mikro2nd 5y ago
    I am forced to wonder, though, whether Norvig's definition is a useful one. By this measure we'd have to consider bacteria to be intelligent. Corporations, too, for that matter. (I do go along with author Charles Stross, occasionally seen here on HN: that Corporations are Old Slow AIs, in which case we've already had AIs around for centuries.) Personally I'm happy to grant a degree of intelligence to plants (and probably even bacteria, if I squint hard enough) though it's of a quite different nature to our own. Certainly feedback loops are central to the idea of intelligence, but there's a whole lot more wanted/needed than merely feedback. And so, I find Norvig's definition a tad wide of the sort of intentionality and sentience we'd want flying a plane, and certainly far, far short of the sort of thing we'd call AGI.
    - dr_dshiv 5y ago
      The discussion is not about AGI. Is it useful to view corporations as old, slow AI? I certainly think so. Otherwise we get really confused about AI. Look at Zillow. That was a deep misunderstanding of AI— the product isn’t done until we take the humans out of the equation. No. What is intelligent is to have a system that uses its own measures of success to improve. This, by the way, is why cybernetics is so critical to understand in the context of AI design.
      - mannykannot 5y ago
        I cannot help but feel that this is just extending the confusion to Zillow. It seems utterly implausible that Zillow's ill-advised zeal for removing people out of the process was driven by a overarching desire to develop AI, as opposed to, say, for making more money.
        dr_dshiv 5y ago
        I would yield to better evidence, but my suspicion is that it stemmed directly from executive-level confusion about AI. Consider: how many investor pitches have you seen that claim “AI” technology as a mechanism to increase perceived IP value? They were telling their investors that they had AI, and AI means people aren’t involved (fallacy).
        mannykannot 5y ago
        No doubt some pitches do claim that AI will increase perceived IP value, but for an investor to go from perceived IP value to the conclusion that people are not involved seems to be a completely unjustified conclusion, and I see no evidence that people are actually thinking this way. Furthermore, I have no idea how you think this line of thought justifies calling 1913's very primitive autopilot technology AI.
        dr_dshiv 5y ago
        I have encountered CEO boardroom thinking that, for instance, suggested that data scientists were not necessary because AI would replace them. I have experience in adaptive education where massively expensive teams of engineers missed the point that the “smartness” of the system needs to be based on improving outcome measures (namely, learning outcomes) and instead focused on massive, complex modeling initiatives with no feedback loops to indicate whether the models were doing anything useful. If more people understood why a steam governor or a 1914 autopilot or a corporate bylaw were primitive forms of AI, they wouldn’t be looking for magic. “If I can understand it, it must not be AI”
  - HPsquared 5y ago
    That definition sounds more like a definition of control systems.
    - dr_dshiv 5y ago
      Note the author alluding to the steam governor as the start of AI. Norvig points to the convergence of control systems, cybernetics and AI
    - fault1 5y ago
      many formulations of RL systems and optimizing feedback control both have origins in for example, in optimal control theory: https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_principle https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_princip...
  - throw0101a 5y ago
    > A feedback loop — namely between a system and its measure of its own performance— is central to the idea of AI. A feedback loop may be necessary for AI (or even "natural" intelligence, and sentience (nevermind sapience)) to happen, but is simply having it sufficient?
    - dr_dshiv 5y ago
      This is a well-phrased question. Is a reward feedback loop sufficient to create intelligence? DeepMind claims “yes” in their 2021 paper “Reward is Enough”: https://deepmind.com/research/publications/2021/Reward-is-Enough https://deepmind.com/research/publications/2021/Reward-is-En...
  - tomrod 5y ago
    Respectfully, while the optimization functions and constraint handling may have overlap, they differ in their intended applications. For optimal control there is no adaptation outside the initial ruleset, which is the core of AI. Optimal control is "keep things on path"
    - fault1 5y ago
      But in many industrial control systems, there is inherent adaptation required to deal with exogenous shocks or noise: https://en.wikipedia.org/wiki/Data-driven_control_system https://en.wikipedia.org/wiki/Data-driven_control_system see also the last few chapters here; http://lavalle.pl/planning/book.html http://lavalle.pl/planning/book.html i would say relative to classical control, fast gpus have replaced the need to have certain closed form solutions that are easily analyzable.
      - tomrod 5y ago
        My working model here: rocket science and mining use integrals, yet are different.
mannykannot 5y ago
It does not help to call that technology AI. If we do so, now we have to invent some new term to distinguish between that sort of "AI", and the sort of AI that could possibly replace pilots, as the former sort obviously cannot.
- dr_dshiv 5y ago
  “Replacements for people” is a terrible definition of AI, in theory and in practice. Alexa, Tesla autopilot, Google search, speech to text, recommendations— any practical example of AI— these are tools. Not human replacers.
  - mannykannot 5y ago
    The author is not defining AI as such; he is proposing it as an application thereof. Is it your position that commercial flying can never be automated, or only that, if it is automated, it would not, by definition, be AI?
    - dr_dshiv 5y ago
      When you say “position that commercial flying can never be automated,” I’m assuming this to mean “fully automated.” My position is that “fully automated” is a logically inconsistent human objective. It means that there is no possibility of human control, because if there is human control / supervision, the system is not fully automated. And, so long as there is human control, you are designing tools for people to use. No one wants anything to be fully automated. It’s the biggest fallacy of AI in design.
      - mannykannot 5y ago
        I wasn't asking about what you think of it as an objective, but whether you think it is possible. Furthermore, when being considered as an objective, whether it is done by anyone's definition of AI seems utterly beside your point. Neither the feasibility of automating the pilot's job, nor whether it is desirable to do so, depends on whether 1st. generation autopilots are considered to be AI.
peter303 5y ago
The term cybernetics was used for both analog and digital feedback automation systems. There is overlap with AI.
drpixie 5y ago
As someone who uses advanced autopilots very regularly, I can inform you that there is no intelligence. Even the best APs use only very basic feedback loops, which are tuned by the aircraft designers and fixed. Their behaviour is not modified by any smart overarching control. A "good" three axis autopilot consists of 3 quite independent systems. One controls pitch/altitude using a feedback loop very similar to an old-style car cruise control (no traffic sensing here :). A second system controls heading, similarly simplistic. And a third controls yaw, even simpler. There might be an auto-throttle, almost identical in function to conventional cruise control. There might be some useful variations (eg VS and FLC modes) - all very useful but there are NO SMARTS HERE. Modern systems have edge-case handling (triggered by too-slow, excess load, or anything going outside the normal operational envelope). The AP handles such cases by going "beep" and turning itself off - just handing over to the pilot!

throw0101a 5y ago

I'm not sure a feedback loop is "clearly" AI: * https://www.electronics-tutorials.ws/systems/negative-feedback.html https://www.electronics-tutorials.ws/systems/negative-feedba... * https://en.wikipedia.org/wiki/Feedback https://en.wikipedia.org/wiki/Feedback There are no heuristics involved: * https://en.wikipedia.org/wiki/Heuristic_evaluation https://en.wikipedia.org/wiki/Heuristic_evaluation * https://en.wikipedia.org/wiki/Heuristic_(computer_science) https://en.wikipedia.org/wiki/Heuristic_(computer_science)

bencoder 5y ago
Seems like an example of the AI effect: https://en.wikipedia.org/wiki/AI_effect https://en.wikipedia.org/wiki/AI_effect although I think when autopilot was invented, the concept of "AI" didn't exist, so maybe not a perfect example
dr_dshiv 5y ago
Autopilot is clearly AI. If I didn’t mention 1914, you wouldn’t doubt it. A feedback loop — namely between a system and its measure of its own performance— is central to the idea of AI. At least according to Peter Norvig, the director of research at Google, who defines intelligence as ‘the ability to select an action that is expected to maximize a performance measure’ (Russell & Norvig, 2016, p. 37).
- mikro2nd 5y ago
  I am forced to wonder, though, whether Norvig's definition is a useful one. By this measure we'd have to consider bacteria to be intelligent. Corporations, too, for that matter. (I do go along with author Charles Stross, occasionally seen here on HN: that Corporations are Old Slow AIs, in which case we've already had AIs around for centuries.) Personally I'm happy to grant a degree of intelligence to plants (and probably even bacteria, if I squint hard enough) though it's of a quite different nature to our own. Certainly feedback loops are central to the idea of intelligence, but there's a whole lot more wanted/needed than merely feedback. And so, I find Norvig's definition a tad wide of the sort of intentionality and sentience we'd want flying a plane, and certainly far, far short of the sort of thing we'd call AGI.
  - dr_dshiv 5y ago
    The discussion is not about AGI. Is it useful to view corporations as old, slow AI? I certainly think so. Otherwise we get really confused about AI. Look at Zillow. That was a deep misunderstanding of AI— the product isn’t done until we take the humans out of the equation. No. What is intelligent is to have a system that uses its own measures of success to improve. This, by the way, is why cybernetics is so critical to understand in the context of AI design.
    - mannykannot 5y ago
      I cannot help but feel that this is just extending the confusion to Zillow. It seems utterly implausible that Zillow's ill-advised zeal for removing people out of the process was driven by a overarching desire to develop AI, as opposed to, say, for making more money.
      - dr_dshiv 5y ago
        I would yield to better evidence, but my suspicion is that it stemmed directly from executive-level confusion about AI. Consider: how many investor pitches have you seen that claim “AI” technology as a mechanism to increase perceived IP value? They were telling their investors that they had AI, and AI means people aren’t involved (fallacy).
        mannykannot 5y ago
        No doubt some pitches do claim that AI will increase perceived IP value, but for an investor to go from perceived IP value to the conclusion that people are not involved seems to be a completely unjustified conclusion, and I see no evidence that people are actually thinking this way. Furthermore, I have no idea how you think this line of thought justifies calling 1913's very primitive autopilot technology AI.
        dr_dshiv 5y ago
        I have encountered CEO boardroom thinking that, for instance, suggested that data scientists were not necessary because AI would replace them. I have experience in adaptive education where massively expensive teams of engineers missed the point that the “smartness” of the system needs to be based on improving outcome measures (namely, learning outcomes) and instead focused on massive, complex modeling initiatives with no feedback loops to indicate whether the models were doing anything useful. If more people understood why a steam governor or a 1914 autopilot or a corporate bylaw were primitive forms of AI, they wouldn’t be looking for magic. “If I can understand it, it must not be AI”
- HPsquared 5y ago
  That definition sounds more like a definition of control systems.
  - dr_dshiv 5y ago
    Note the author alluding to the steam governor as the start of AI. Norvig points to the convergence of control systems, cybernetics and AI
  - fault1 5y ago
    many formulations of RL systems and optimizing feedback control both have origins in for example, in optimal control theory: https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_principle https://en.wikipedia.org/wiki/Pontryagin%27s_maximum_princip...
- throw0101a 5y ago
  > A feedback loop — namely between a system and its measure of its own performance— is central to the idea of AI. A feedback loop may be necessary for AI (or even "natural" intelligence, and sentience (nevermind sapience)) to happen, but is simply having it sufficient?
  - dr_dshiv 5y ago
    This is a well-phrased question. Is a reward feedback loop sufficient to create intelligence? DeepMind claims “yes” in their 2021 paper “Reward is Enough”: https://deepmind.com/research/publications/2021/Reward-is-Enough https://deepmind.com/research/publications/2021/Reward-is-En...
- tomrod 5y ago
  Respectfully, while the optimization functions and constraint handling may have overlap, they differ in their intended applications. For optimal control there is no adaptation outside the initial ruleset, which is the core of AI. Optimal control is "keep things on path"
  - fault1 5y ago
    But in many industrial control systems, there is inherent adaptation required to deal with exogenous shocks or noise: https://en.wikipedia.org/wiki/Data-driven_control_system https://en.wikipedia.org/wiki/Data-driven_control_system see also the last few chapters here; http://lavalle.pl/planning/book.html http://lavalle.pl/planning/book.html i would say relative to classical control, fast gpus have replaced the need to have certain closed form solutions that are easily analyzable.
    - tomrod 5y ago
      My working model here: rocket science and mining use integrals, yet are different.

mannykannot 5y ago

It does not help to call that technology AI. If we do so, now we have to invent some new term to distinguish between that sort of "AI", and the sort of AI that could possibly replace pilots, as the former sort obviously cannot.

dr_dshiv 5y ago
“Replacements for people” is a terrible definition of AI, in theory and in practice. Alexa, Tesla autopilot, Google search, speech to text, recommendations— any practical example of AI— these are tools. Not human replacers.
- mannykannot 5y ago
  The author is not defining AI as such; he is proposing it as an application thereof. Is it your position that commercial flying can never be automated, or only that, if it is automated, it would not, by definition, be AI?
  - dr_dshiv 5y ago
    When you say “position that commercial flying can never be automated,” I’m assuming this to mean “fully automated.” My position is that “fully automated” is a logically inconsistent human objective. It means that there is no possibility of human control, because if there is human control / supervision, the system is not fully automated. And, so long as there is human control, you are designing tools for people to use. No one wants anything to be fully automated. It’s the biggest fallacy of AI in design.
    - mannykannot 5y ago
      I wasn't asking about what you think of it as an objective, but whether you think it is possible. Furthermore, when being considered as an objective, whether it is done by anyone's definition of AI seems utterly beside your point. Neither the feasibility of automating the pilot's job, nor whether it is desirable to do so, depends on whether 1st. generation autopilots are considered to be AI.

peter303 5y ago

The term cybernetics was used for both analog and digital feedback automation systems. There is overlap with AI.

drpixie 5y ago

As someone who uses advanced autopilots very regularly, I can inform you that there is no intelligence. Even the best APs use only very basic feedback loops, which are tuned by the aircraft designers and fixed. Their behaviour is not modified by any smart overarching control. A "good" three axis autopilot consists of 3 quite independent systems. One controls pitch/altitude using a feedback loop very similar to an old-style car cruise control (no traffic sensing here :). A second system controls heading, similarly simplistic. And a third controls yaw, even simpler. There might be an auto-throttle, almost identical in function to conventional cruise control. There might be some useful variations (eg VS and FLC modes) - all very useful but there are NO SMARTS HERE. Modern systems have edge-case handling (triggered by too-slow, excess load, or anything going outside the normal operational envelope). The AP handles such cases by going "beep" and turning itself off - just handing over to the pilot!