Do companies here think people don’t really understand what “AI” means?

Was watching TV and saw an ad for a Mitsubish AC which said that the AI will detect where the temperature differs in the room to send warm/cold air that way. That’s not AI… that’s just a sensor.
Also, watching 紅白歌合戦, they said they used AI to bring a younger version of a singer back. It was just a CGI version of the singer, singing next to her real-life self.

I thought that “AI” has something to do with learning or creating something automatically. Or am I wrong?

23 comments
  1. Both can still be true, they could have used an AI to help the sensor, could’ve use an AI to help do the rendering. I could even use an AI to do that comment! Don’t need it, but can do it.

  2. I personally think the whole world has jumped the gun on this one.
    So much AI these days is just artificial automation. AI was supposed to be autonomous, unpredictable and versatile. An algorithm feeding me content based on clicks is not AI.

    To answer your question there may be “AI” somewhere in there. The algorithm to make the cg may have gathered the info intellegently. The aircon may learn how sending airflow in one direction creates cold spots and use machine learning to do better and better?

  3. AI doesn’t have to do with learning — that’s Machine Learning, a subset of AI.

    By a classical definition, a system that detect temperature and adjust the air temperature is an AI — usually a rule-based AI.

  4. AI is much broader than just machine learning. A simple algorithm to play rock-paper-scissor can be classified as an AI.

    There is a quote I like: “As soon as it works, no one calls it AI any more”.

  5. It’s artificial and it’s intelligent. It’s AI.
    Unless you think smart home sensors are apart because they’re _really_ smart.

    The thing about AI is that the term is literally meaningless. Maybe you’re thinking about machine learning when you use the term AI but that’s a _you_ problem, not a _Japan_ problem.

  6. While it’s not exactly accurate, current AI-marketed products usually refer to the technology being used, specifically deep learning. If the product does employ trained deep learning models, manufacturers will market it as such.

  7. My spidey sense tells me its a somewhat nuanced marketing gimmick, much like SDG, ESG, gigabit, smart, etc.

    Is it AI? Could be. Doesn’t matter, though, because the companies likely won’t be met with many if any calls to clarify, but still reap the rewards that I guess they’re getting by playing the AI card.

    What I can say is that, anecdotally, there’s a conspicuously low level of use or access to public AI resources in the particular corners of the industries I operate in: IT, entertainment, and real estate. In my experience, this slow build to engagement is usually due to a combination of a lack of localized materials, unfamiliar UX/UI, and organizational hesitancy towards new technologies.

    Which isn’t to say Japan isn’t doing anything with AI – there are researchers and developers for sure – but from the point of view of The People, as it were, I don’t see them being provided with the means to dig in.

  8. AI = Artificial Intelligence

    It can literally be anything. Including a sensor with an if statement to say if it gets above X do Y else do Z.

  9. THEY (the business and government greedbags) don’t know what AI means. And they don’t care either.

    Its same thing as when they label things as “green” such as waste incinerators.

    They just want the image and/or the sales. Words like integrity, honesty and responsibility also have no meaning to them….just words to be exploited.

  10. AI just makes your device smarter, the sensor collects data and they could run it through a model to adjust temperature in a smarter way. AI is very broad , it can be simple as if temperature above a threshold, switch AC on. Or as complex as using a variety of sensor data such as humidity , temperature, camera sensors to adjust room climate accordingly.

  11. > Also, watching 紅白歌合戦, they said they used AI to bring a younger version of a singer back. It was just a CGI version of the singer, singing next to her real-life self.

    FYI I’ve seen a lot of demonstrations where they feed the computer footage of someone, and out comes a 3D model that will say and do whatever you want. My guess is that they used this tech, at least to a certain extent.

  12. Technical terms are routinely misused in advertising/marketing. It’s not the techies writing the copy.

  13. Well technically there is no true AI yet, so we’re semi-forced to broaden the term.

    Basically even the best AI right now still follows the preprogrammed: “IF A, DO B” command. So yes, sense temperature and changing temp to preset is the best we got right now to AI.

    And this isn’t just Japan. It’s same globally. Anything that says AI in your home country is also technically bs.

  14. >Also, watching 紅白歌合戦, they said they used AI to bring a younger version of a singer back. It was just a CGI version of the singer, singing next to her real-life self.

    My understanding from the introductions to the performance: they used AI to compile data from Matsutoya Yumi’s past performances and recordings over the years to train the AI to simulate her voice when she was younger (her vocal quality sounds pretty different today from back then) to sing a new song (a previously unreleased track with rewritten lyrics) with her present, older self. The CGI was just for the presentation, the AI refers to the vocal (re)production.

  15. AI doesn’t have as much to do with detection as it does with an algorithm generating behavior based on information it collects, especially when said information is used by the algorithm to form ever-more-accurate guesses for the “right answer” humans expect.

    In the case of an AC unit that detects temperature and adjusts itself to address a particular hot spot, the AI part is not the sensor, but rather the algorithm that analyses the information coming in from the sensor and generates a response. Based on how the algorithm has learned from previous guesses on how to address hot spots and using the sensor information, it makes a new guess of how to address this particular hot spot. If it receives positive feedback, the algorithm learns that the guess was a good one, and it will use that information for future guesses.

    In the case of generating images, the AI is not in the projection of the image but rather in how the image was/is generated. In the case of humans (and especially with realistically animated human images), the AI algorithm references countless other images of humans to render what it guesses is an accurate representation of the target person and to accurately animate that image with realistic physics and expressions that correspond with human anatomy in general and with the target person in particular.

  16. Yes and no. Most AI here in Japan is just rules based software. But given how un-automated many day to day things have been in the country until recently, any new software seems to be like a miracle blackbox that makes things better, and slapping the label AI on it is cool.

    The problem comes from, and this is the misunderstanding point, is many confuse these complex, but non-intelligent software, as AI, and assume they are actually doing some more than they are…. that they have some type of learning back end, or dynamic decision making, when they don’t. They are just VERY complete rule sets with no ability to deviate from their original programming or adapt in any meaningful way.

  17. The birth of the term AI was to answer the question of “can machines think like human or have human insight”. First machine learning goes back 1952 and its only an AI for computer game of checkers.

    In current era, AI can recognize face, hear and talk, generate art, deep learning. More complex algorithms and even smarter than human in its specially designed task.

    IMO people should stop calling simple device with “simple task” an AI. They should be called automated machines instead. Or maybe classify them in different term based on how much computing they do or how modern their algorithms are.

    Unfortunately we can’t stop companies. Because attaching AI is cool and it brings profit. The marketing department like it.

  18. While there are so many misuse of the word around, there are already software that can replicate someone’s voice using machine learning and such. You could teach an AI with some voice recordings and then ask it to replicate it. There are already consumer software that would do this.

  19. There’s AI when I watch sports match on the tube. One team scores and the AI updates the score on the screen.

Leave a Reply
You May Also Like