back to article Vendors rush to call everything AI even if it isn't, or doesn't help

Many enterprise software vendors “are focused on the goal of simply building and marketing an AI-based product rather than identifying use cases and the business value to customers.” So says Gartner analyst Jim Hare in a July 6th piece of research titled “How Enterprise Software Providers Should (and Should Not) Exploit the AI …

  1. Anonymous Coward
    Anonymous Coward

    AI exists?

    When I can ask a device to complete a task it wasn't specifically programmed to do then I'll believe AI exists. Until then an algorithm that analysis cat photos and does nothing else than churn out predetermined results is a poor attempt at AI.

    1. Anonymous Coward
      Anonymous Coward

      Re: AI exists?

      "When I can ask a device to complete a task it wasn't specifically programmed to do then I'll believe AI exists."

      You may be interested to read the wealth of literature on the topic then, because that's *exactly* what the state of the art does. My particular favourite is the 2016 paper entitled "Hair Segmentation Using Heuristically-Trained Neural Networks".

      1. BillG
        Megaphone

        Re: AI exists?

        I think the issue with choosing AI systems is that there are no meaningful metrics or benchmarks for comparing different products.

      2. Anonymous Coward
        Devil

        Re: AI exists?

        > When I can ask a device to complete a task it wasn't specifically programmed to do then I'll believe AI exists

        > [...]that's exactly what the state of the art does:

        For example, the R2D2 knock-off robot that fell down the stairs into a pond at some Mall.

    2. JeffyPoooh
      Pint

      Re: AI exists?

      Those that are amazed by the present state of the art in AI must have a very low threshold of amazement. There's not been any significant, discreet, newsworthy 'breakthrough' at all; just the inevitable march of Moore's Law and minor algorithm tweaks. An AI researcher from the 1960s could be caught-up in about 15 minutes.

      "Ooooohhh look, virtual neural nets with layers, so many of them, and an executive to manage the learning. So EXCITING..."

      Yawn.

      1. Anonymous Coward
        Anonymous Coward

        Re: AI exists?

        "There's not been any significant, discreet, newsworthy 'breakthrough' at all"

        Hinton 2006 leading to Krizhevsky in 2012 leading to the victory of a computer over humans in Go, a feat not that long ago thought completely and utterly impossible.

        Discrete breakthrough (you meant discrete?) leading to practical application leading to previously unattainable innovation.

        Just because you dislike AI/ML marketing does not mean "narrow" AI using modern ML techniques is not a very real, very effective thing.

    3. Anonymous Coward
      Joke

      @Mihto

      "When I can ask a device to complete a task it wasn't specifically programmed to do then I'll believe AI exists."

      The Tesla (and Google, but I like Tesla better as target) were never programmed to crash into other cars yet it still happened. Oh dear: AI confirmed, Musk was right: they're out to kill us! ;)

  2. Anonymous Coward
    Anonymous Coward

    Wow

    An analyst who actually understands what is going on. Kudos!

    1. John Brown (no body) Silver badge
      Thumb Up

      Re: Wow

      From the article: "Gartner analyst Jim Hare in a July 6th piece of research"

      I guess he's been reading the El Reg comments section over the past years worth of AI stories. No other "research" required since so many of us have called this so many times over that time span.

      1. Anonymous Coward
        Anonymous Coward

        Re: Wow

        To be fair, many of us were calling them wrong when they kept saying PC sales would rebound after they first dipped in 2011. They got that wrong for years, so if they've learned to take commentards thoughts into account before publishing that's new! :)

    2. Captain DaFt

      Re: Wow

      An analyst who actually understands what is going on. Kudos!

      And a Gartner analyst, no less! I'm agog!

      1. Chemical Bob
        Terminator

        Re: Wow

        Must be a bot, no real human at Gartner has ever been able to recognize the obvious.

  3. DropBear
    Facepalm

    "heading toward"...? "likely to become"?!? Sorry, that happened at least a year or two ago...

  4. Mage Silver badge
    Coffee/keyboard

    We are heading toward

    No, we got there years ago when they started calling "Expert Systems", "Pattern Recognition" and big human curated databases with funky "natural" search stuff like "Cognitive" or "AI".

    There is no AI in the sense that even 1980s Computer Science researchers hoped for in the 1980s. There is still AI research, though most of it isn't.

    1. Anonymous Coward
      Anonymous Coward

      Re: We are heading toward

      Modern ML systems, many of which are labelled AI, bear absolutely no resemblance to expert systems or to classical pattern recognition systems. Yes, all these things take some input and make a decision about the output, but the way in which the decision is made and (more importantly) the way the decision mechanism itself is formulated could not be more different.

      1. TRT Silver badge

        Re: We are heading toward

        Do you remember "Fuzzy logic"? That genius buzzword that sold us washing machines and microwaves etc. that were apparently designed to cope with more than one sensor input at a time - extending wash times if the water temperature was lower, or giving it an extra rinse if the water was highly conductive due to a lot of dissolved particles (meaning, presumably, that the clothes were super dirty). I wonder whatever happened to those?

      2. Anonymous Coward
        Anonymous Coward

        Re: We are heading toward

        So ML systems are more advanced than the expert systems of the 90s? It would be sad if that wasn't the case!

        Still doesn't mean today's ML systems are AI. They are not, not even close. We have a bunch of idiot savant programs that can handle things they're taught to do (chess, Go, speech recognition, etc.) but display not even the tiniest bit of general intelligence. When a Go program can play chess by being told the rules of the game and watching people play, or a speech recognition program can learn to recognize handwriting on its own without being explicitly programmed for the task, then I'll be impressed.

        1. Anonymous Coward
          Anonymous Coward

          Re: We are heading toward

          "So ML systems are more advanced than the expert systems of the 90s?"

          Who said anything about advanced? Plenty of ML systems fucking suck. See: "Watson". I said they're radically different in design, construction and operation. Comparing the two is like comparing a search algorithm to a sort algorithm. Sure you could use them to do the same thing, but they aren't the same thing.

          "When a Go program can play chess by being told the rules of the game and watching people play..."

          You know that's exactly, literally how AlphaGo works right?

          Well, not exactly. It spent most of its time teaching itself how to play the game by playing multiple versions of itself. Humans were too irrational; AlphaGo was instructed to optimise for chance of victory. Humans can try, but they can't always do that, so learning based on the human history of the game reached a cap in performance below that of the top players.

          "a speech recognition program can learn to recognize handwriting on its own "

          Given that's impossible even for humans (see: Rosetta Stone) that would indeed be seriously impressive.

          1. Anonymous Coward
            Anonymous Coward

            Re: We are heading toward

            AlphaGo was first programmed with basic knowledge of Go, it played itself to improve but it didn't start from a blank slate.

  5. Mage Silver badge
    Joke

    Hare says users are already seeing through the hype and just not buying AI

    Such honesty. Will his company let him write public reports again?

    We need more such frank analysis. I'm looking at you Analysys Mason, a consultancy company that helps market mobile products for Infrastructure and mobile providers.

    Or Gartner generally.

    1. Anonymous Coward
      Anonymous Coward

      Re: Hare says users are already seeing through the hype and just not buying AI

      He's so different from the regular Gartner analysts it makes me wonder if he's an AI!

  6. Aristotles slow and dimwitted horse

    Hype + Marketing

    Whoda' thunk it...

  7. Doctor Syntax Silver badge

    "marketers have brainwashed themselves"

    Depends on a fact not in evidence as Perry Mason used to say.

    1. Chemical Bob
      Gimp

      Well, they've washed *something*!

  8. Sanguma

    there's always Natural Stupidity

    but the companies concerned are already exploiting that with their over-hyped advertising and marketing ...

    speaking facetiously, of course - I can remember the days when BYTE Mag proudly boasted ads proudly boasting Desktop Mainframes containing 32-bit i80486 processors, wow!!! Not a word about 99.999% reliability, failover, the like ...

  9. Pascal Monett Silver badge
    Thumb Down

    "his document predicts it will be pervasive by 2020"

    Fat chance of that. For AI to be pervasive, it first has to exist.

    It doesn't. End of story.

  10. Anonymous Coward
    Anonymous Coward

    I'm contemplating putting together a commentard bingo card for any story related to AI or ML, including such gems of insight like:

    "It's just an expert system!"

    "Cat pictures"

    "We were doing this in the seventies"

    "If it can't have a conversation it's not REAL AI"

    etc. etc.

    The problem is I think we'd get a winner just too damned quickly.

    1. TRT Silver badge

      The Great Hyperlobic Omni-Cognate Neutron Wrangler...

      could talk all four legs off an Arcturan MegaDonkey - but only I could persuade it to go for a walk afterwards.

  11. Terry 6 Silver badge

    over-promise, under-deliver

    Marketing types, and to be honest I don't think it's their fault* seem to build their lives on that premise.

    *My impression is that they are merely doing what they need to do to keep their jobs under pressure from bosses who say things along the lines of "We need to show we're already in (Target_name=latest new thing)". And I'm guessing that they in turn are worried about share holders, who are fickle and prone to sell established and buy trendy-new stocks.

    1. FlossyThePig

      Re: over-promise, under-deliver

      Can we add "agile" and "cloud" to the list

      1. jake Silver badge

        Re: over-promise, under-deliver

        And DevOps.

  12. Third Man
    IT Angle

    Not the worst way to destroy excess value...

    Wars are worse - marketing hype kills fewer - maybe?

    1. John Brown (no body) Silver badge

      Re: Not the worst way to destroy excess value...

      "marketing hype kills fewer - maybe?"

      Depends on how you define "marketing". I'm sure the arms industry does marketing.

  13. Anonymous Coward
    Anonymous Coward

    AIIII... LMAO!

    Avoiding Another AI Winter, James Hendler, in: IEEE Expert, March/April 2008

    He forgot "Vendor overhype"

  14. Alister
    Holmes

    Gartner Analysis = "No shit, Sherlock"

    1. Ottman001

      That it is but I'm sure we've all met people who need the bleeding obvious pointed out to them.

  15. Shameless Oracle Flack
    Stop

    Great piece, it's more about platforms that deliver broad capabilities than about specific new shiny AI things.

  16. I Like Heckling Silver badge

    How is this different to the beginning of the 2000's when every manufacturer rushed out and slapped a tiny little, barely readable LCD onto anything and everything they could and boasted that it was now DIGITAL.

    It's time to end this kinda of trumped up marketing bullshit that only serves to deliberately misdirect and mislead consumers. Marketing 'spin' is getting worse than government 'spin' these days... but at least marketers aren't claiming success for their handling of some disaster that could have been avoided and peoples lives saved if they'd not been responsible for the cuts that ensured said disaster in the first place.

    1. jake Silver badge

      Beginning of the 2000s?

      That was the mid 1970s ... Neighbor across the street from my parents bought a new mailbox with a "programmable" LED display for the house number. Ran off the low-voltage doorbell circuit (new LED doorbell button included!!!). The user's manual was plastered with the word "DIGITAL!", in all CAPS with the exclamation point ... I was asked to "program" the silly thing for him. For values of "programming" that included placing jumpers properly ... Only five digits, they didn't even scroll. But it was DIGITAL! with a low, low cost $499.95, in 1975 money. At least it wasn't battery powered ...

      And of course, during that decade, we had "digital" alarm clocks, watches, microwave ovens, televisions, VOMs, indoor/outdoor thermometers and the like, all of which were "digital" only in display. But we were going DIGITAL!, if you believed the marketing hype. Sound familiar?

  17. JeffyPoooh
    Pint

    Same thing...

    Vendors rush to call everything "3D Printing" even if it isn't, or doesn't help

    There was an example where they literally plastered up a new sign "3D Printing" on an existing factory.

    1. John 104

      Re: Same thing...

      I may be dating myself here, but this was the same sort of thing with VCRs back in the 90s.

      Everything was advertised as HQ. "4 head vcr with HQ" was the hype. Labels on all the boxes, panels, everywhere. What was this amazing HQ they spoke of? High Quality. Fucking genius.

      1. jake Silver badge

        Re: Same thing...

        My soon-to-become-wife pronounced it "Howse Qwap".

  18. jake Silver badge

    My large animal Vet came in the other day.

    This guy's in his second career ... he became a Vet after 25 years as a DBA working for IBM. He knows I'm a computer guy, and thought I'd be amused by a bit of advertising he had received. It was for a large animal veterinary practice management software package "NOW WITH AI!

    The Vet was laughing, and wondered how many times the company in question got Vets inquiring about their new Artificial Insemination package. Without a pause, I dialed the 800 number ... the answer was over 80% of calls! The guy on the other end wasn't amused when I suggested they fire their marketing genius and hire an AI expert ...

  19. Barry Rueger

    Don't Bother Me With Details

    I'm pretty up to date with all of my computer lingo, and know my way from a modem, to a Zip Drive, to The Cloud, but I can't think of any use case where I would be likely to say "Does it have AI?"

    I'm a lot more interested in what, exactly, a product does that is useful to me. If you're flogging AI you better be able to explain what that actually does that makes your (probably more expensive) product better than what I'm already using.

    Sure it's nice to have software that can make good guesses about what you might need next, but to date none the AI labelled products seem to take that much beyond what various music recommendation websites were doing in the 90s.

  20. Chemical Bob

    applying the AI label too indiscriminately

    Question for the Gartner analyst:

    What, exactly, is the correct amount of "applying indiscriminately"?

  21. Alan Bourke

    You could knock a Hangman game on the ZX Spectrum

    and pass it off as AI these days, it would seem.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like