Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
×
AI Microsoft Science

AI-Enabled Cheetos Offer Promise of the Perfect Puff (wsj.com) 66

Posted by BeauHD from the let's-get-scientific dept.
Microsoft says in a blog post that PepsiCo is using their Project Bonsai "machine teaching" service to "help ensure its Cheetos cheese-puff snacks all have the same texture, crunch and shape," reports The Wall Street Journal. From the blog post: PepsiCo built a computer vision system that continually monitors Cheeto attributes. Data about qualities such as density and length are fed to the Project Bonsai solution, which makes adjustments to bring the product within spec. This approach reduces the time it takes to correct inconsistencies and allows operators to focus on parts of the line that require human expertise. PepsiCo is preparing to use the solution in a production plant and exploring how to use the solution with other products, including the tortilla chip manufacturing process. An out-of-spec product can't be sold, which leads to wasted resources, time, and money. Greater consistency helps PepsiCo maintain high quality products while maximizing throughput.

To make an ideal Cheeto, the solution needed examples of what wasn't ideal -- and needed to know what to do in those cases. The extruder line is self-contained and well-suited for developing and testing an autonomous system solution. Operators had been running it manually, which gave developers the opportunity to build the solution from scratch, instead of on top of other software. The AI solution has a recommendation mode and a closed loop control mode. In both modes, a computer vision system continuously measures the quality of the Cheetos. In recommendation mode, the AI will alert an operator if the product drifts out of spec, displaying on an instrument panel the attributes that are not ideal as well as a recommendation to correct it. The operator can push a button to make any or all recommended adjustments.

In control mode, the only difference is that the AI solution skips the recommendation step and adjusts the extruder line specifications independently. The company expects that running this intelligent control system will return product to acceptable attributes faster. In the current extruder line, operators measure product attributes manually at defined intervals. If the Cheetos are out of spec, the operator makes adjustments based on guidelines or experience to return the product to acceptable quality. The problem: Infrequent sampling meant that the line could be producing out-of-spec Cheetos for a longer period of time without anyone realizing. The Project Bonsai solution will monitor the product almost continuously, using sensors to oversee characteristics such as length and bulk density. That way, it knows as soon as the product strays outside a defined range.
This discussion has been archived. No new comments can be posted.

AI-Enabled Cheetos Offer Promise of the Perfect Puff More

AI-Enabled Cheetos Offer Promise of the Perfect Puff

Comments Filter:

  • We need the perfect orange

  • We'll have the same homogenized appearance, texture, and taste. No anomalies or variance allowed.

    How dull and boring.

    • On the other hand, were you ever pleasantly surprised by a noticeably different Cheeto?

      • I don't eat Cheetos (or haven't in who knows how many decades), but having a variety of sizes, tastes and textures is a good thing. Your palate becomes numb (so to speak) if the same food is eaten over and over with no break in between. This article on wine tasting [foodandwin...hetics.com] describes the effect.

        Finding a crunchy Cheeto, or one slightly more or slightly less flavorful, one larger or smaller than the other, is a good thing. As someone below said, all they're doing is creating a cheesy Pringle.

        • Re: (Score:2)

          by Strider- ( 39683 )

          This is why when I eat this kind of junk food, I like the stuff that's the party mix of pretzels, cheetoes, nacho chips, and a few other things. Variety is the spice of life.

    • We'll have the same homogenized appearance, texture, and taste. No anomalies or variance allowed.

      How dull and boring.

      Like the Pringles of cheezies.

    • This isn't about keeping each Cheeto identical (though puffs are already close to that), this keeps them within acceptable tolerance: Some big, some small, but none so dense as to be hard to eat. None so cheese-less that they have to throw them out. Keep in mind: The factory already throws all of the out-of-spec ones out. For people, variety is a great thing. For products, not so much.

      • I'd be surprised if they throw the out of spec ones out, as opposed to them ending up in a bag of dollar store mixed snacks along with the broken chips and pretzels, or in a no name cheezie bin at Bulk Barn.

  • 1. There is NO AI!
    2. If there were AI, this would be the stupidest possible use for it.
    3. There is NO AI!
    I know I repeated 1 and 3, but I felt it was an important enough fact to need repeating.

    • Yes, there is clearly AI. Your definition of AI does not suit this technology, however the rest of the world has moved on without you, as your definition is useless/impractical.

      And no, monitoring the quality of a product to decrease defects and increase saleable yield is not a stupid use of AI.

      ~D

      • No, your definition of AI has nothing to do with artificial intelligence! You are deliberately abusing the term to mean very very shitty simulations that are as close to real neural nets as a perfect sphere on a sinusoidal trajectory is to a horse race (not even exaggerating), and are nothing more than simple vector mappings with weight matrices. Which have the far more mundane name "universal functions". But that won't let you rip off your financers, now would it?

        By you definition, matrix transforms of vec

        • Actually, games are a great example of what I'm talking about when I mention AI. Clearly, there are no artificially intelligent consciousnesses running the opponents in a game. But that distinction isn't helpful to the rest of the world. Yes, it's important to the game developer, but to the game reviewer? They don't care about the implementation. When they review the game for the public, they're not going to write "The finite state automata driving the animation is poorly thought through"... They're going t

        • Re: (Score:2)

          by Merk42 ( 1906718 )

          No, your definition of AI has nothing to do with artificial intelligence! You are deliberately abusing the term to mean very very shitty simulations....

          Unless there is fecal matter in those simulations, well I guess you just used the word "shit" wrong, so now I'm very angry and will go on a diatribe about it!!

    • 4. Even if there were AI, despite the headline - we would still not be talking about AI-Enabled Cheetos. The purported AI is part of the manufacturing process, not the Cheeto itself.

      I mean good grief, is the Cheeto going to scream when you bite down on it?

      • I mean good grief, is the Cheeto going to scream when you bite down on it?

        I just had a genius idea for a product. It’s time the world had some deep fried learning. So, anyone make edible AI chipsets?

    • Re: (Score:2)

      by ljw1004 ( 764174 )

      There is NO AI!

      Oh FFS, "AI" is an industry standard term for machine learning systems. "Neural nets" is another industry standard term. They're not really neural, and they're not really learning, they're just adjusting weights in matrices at huge scale. But those are what the standard terms are.

      And if you think "AI" or "learning" or "neural" are bad words for them because they don't relate to the narrow subset of behaviors that are (really and rightfully) called intelligence, you're just a curmudgeon.

      • You stole the term for your shitty marketing purposes, and now have the audacity to act like you're the normal ones too??

        No, it is a venture-capitalist-rip-off-industry standard term for matrix mutiplication functions used for pattern matching that really really REALLY has fucking nothing to do with artificial intelligence, and barely even with neural nets.

        By your logic, I do "AI" when I transform my 3D model's vertices with a matrix in my game engine. Which goes to ahow how wrong that definition is.

        • It seems you don't understand the difference between a mathematical construct that can hold and process information and the information stored within it. Neural networks are a mathematical approximation to biological neural systems, an extraction of the fundamental mathematical behaviour, without the biologically imperfect representation of those rules. True, some of the current networks miss certain functions, however they can approximate them. This was been mathematically proven back in the 80s (though pr

    • Re: (Score:2)

      by Ranbot ( 2648297 )

      Mod +1 Pedantic

  • Can you imagine even the grandest emperors of ancient china demanding that each grain of rice be hand-selected?

    Yet we eat that now:

    https://www.youtube.com/watch?... [youtube.com]

    • Can you imagine even the grandest emperors of ancient china demanding that each grain of rice be hand-selected?

      Yes, actually, I can picture that quite readily. Historically, monarchs have made all sorts of ridiculous, exorbitant demands.

  • Thank you PepsiCo. You could have reduced your massive output of plastic waste but you know where your consumers' interests and motivations truly dwell. I, for one, miss those potato crisps with the browned edges. They are so rare now due to precisely this approach. But I'm in the minority. Most people seek perfection and this is what PepsiCo seek too. A perfect Cheeto. Just typing that in brings me such joy. I look forward to Dominos and McDonald's utilising the same technology.

  • I wonder how the AI will react if... (Score:1)

    by Anonymous Coward

    I wonder how the AI will react if you feed it pictures of Trump. "It's too puffy, it's too orange, can't correct, abort, Abort!"

  • The AI needs to first start with one important step: All Cheetos Crunchy's should be incinerator fuel in a power station that's used in a nightmare generator for whoever invented the Cheetos Crunchy.

  • is this spoof or real? (Score:3)

    by ljw1004 ( 764174 ) on Thursday December 17, 2020 @08:29PM (#60843324)

    I've read the article through twice. I still can't tell if it's serious or deadpan spoof.

  • How this "AI" approach differs from simple closed-loop feedback that has been in use for the last 60 or so years?
    • In training mode, it is more or less the same, only the AI is being trained to take the operator's job. When done, the operator has no job.
    • Duh, it’s a feed forward control loop.

    • Re: (Score:2)

      by WDot ( 1286728 )
      I do AI, but not controls. My guess is that it IS a closed-loop feedback system, but the difference is the "feedback" is an image from a camera that is fed into an AI which outputs the attributes used to adjust the feedback system (or even more directly, outputs the adjustments themselves). When someone say "AI" don't think "is this an actual sentient conciousness" think "a statistical model that will attempt to map a set of training inputs to training outputs". If you have an algorithm that can reliably do
  • Wasn't the whole point of Krispy Kreme that they were, for all intents and purposes, identical? Not great (tried one back in the day, wasn't impressed) but literally identical?
    Personally I prefer hand-made 'craft' cheetos. The flaws and variations make them more interesting. /s
    In all seriousness why the hell do you need a freakin' 'AI' to do this anyway? Krispy Kreme made doughnuts with an automated machine to get them exactly the same. You're actually saying that they can't make a simple mechanical Cheet

    • Wasn't the whole point of Krispy Kreme that they were, for all intents and purposes, identical? Not great (tried one back in the day, wasn't impressed) but literally identical?

      Personally I prefer hand-made 'craft' cheetos. The flaws and variations make them more interesting. /s

      In all seriousness why the hell do you need a freakin' 'AI' to do this anyway? Krispy Kreme made doughnuts with an automated machine to get them exactly the same. You're actually saying that they can't make a simple mechanical Cheetos machine that makes them identical? This has got to be a joke.

      Krispy Kremes directly out of their fryers were delicious. Krispy Kremes sitting on a gas station shelf for a couple days, not so much.

      Are you asking for the "Artisan" Cheetos?

      As an orange dust delivery system, they are mostly effective.

      I don't ever remember any time in my life thinking "I could go for some Cheetos right now!"

  • Its as much AI as a thermostat, but it's make me Hungry!

  • comes out of my pot pipe. Frequently.

  • Well then, let ME be the first one to welcome our new Cheetos wielding AI overlords, long may they crunch.

    (I'm typing with Cheeto Dust on my fingers right now. it's the same as Dune's Spice IMHO, ignoring those minor pesky differences.)

  • Best (Score:2)

    by Plugh ( 27537 )
    The best cheetos are the deformed ones with much less puff and way more chemical cheese flavoring than the designers planned for

  • OK, the crunch is critical. Having just the right texture in every bite (not byte) just right.
    Americans may remember back to a time when marketers learned to sell the 'sizzle' rather than the steak. It's the same principle. Remember, Pepsi sells sugar water. Marketing is everything when you sell such a basic commodity. They are right to emphasize that interesting texture in their basic puffed corn product.

    Unfortunately, more is required in this case:
    1 The fake cheese flavor. Gawd, that sucks!
    2 The food colo

  • Why don't they use their shitty weight matrices to find out how to turn Cheetos into something that at least comes anywhere close to being food? You know ... for humans.

  • Why now? I've never had a bad Cheeto. Ever.

  • They should train it on the crunch sound, not the way it looks. And maybe on the amount of powder that rises when crushed? You have to like Cheetos a LOT to train this corpus..

  • We're in deep shit now, Francis. They've adapted to the perfect puff. Next thermonuclear warfare, where does the madness end?!?!?!

  • ... the silliest 21st century post so far this century. Is each Cheeto AI-enabled?

Slashdot Top Deals

Any program which runs right is obsolete.

Close