This is a problem that requires zero AI.
A touchscreen kiosk is all you need to have the customer place their own order. You just need to have 2-3 ordering stations as they will be slower than trainer staff.
But shareholders demand AI
This is a problem that requires zero AI.
A touchscreen kiosk is all you need to have the customer place their own order. You just need to have 2-3 ordering stations as they will be slower than trainer staff.
But shareholders demand AI
Since 2023, the fast-food chain has introduced the technology at over 500 locations in the US, with the aim of reducing mistakes and speeding up orders.
Bullshit, they just want to save money by replacing employees with AI.
Step 1: replace all your workers with AI Step 2: set new record for unemployment Step 3: most people can't afford your product Step 4: increase the price to compensate for decreasing sales Step 5: blame Obama
Step 1: Replace your executive and board members with LLMs. Despite LLMs being increadibly unintelligent, there will be no loss of leadership IQ in the corporation because they were already unintelligent enough to put a LLM where it doesn't belong because they don't understand what it is and what it isn't.
Step 2: Enjoy the savings of those large useless eater salaries you no longer have to pay.
They don't even reduce cost from the savings. They charge the exact same amount until they want to increase prices.
Oh shit, I didn't even notice that part. Yup, that's a whopper, and not the kind you eat.
This is not AI failing to do an easy job. This is "unskilled" labor doing complex and demanding work that cannot be duplicated by trillion dollar software.
I mean, unskilled just means minimal extra training is needed, not that it's not complicated. Actual non-complicated jobs were automated last century in the West.
Tbh this is an incredibly easy fix, either cap the number of waters someone can order in software or have an override where a human takes over if an order is suspicious, there's not an infinite number of ways to fuck with this.
Capping waters fixes that one specific issue but not the problem.
A suspicious order isn't easy to define and no person who has ever participated in software development would underestimate the infinite ways a User can break software.
Surely if the person making the order sees 18,000 waters they would think, hold on this doesn’t seem right maybe I should ask the customer if they really want 18,000 waters?
The same applies for the ice cream with bacon on it which was mentioned in the article. I believe a lot of these could be resolved with a bit of common sense.
The same applies for the ice cream with bacon on it
Does it, though? Unlike the 18,000 waters, if I were working a drive through I wouldn't even blink at an order for bacon ice cream. Heck, I might make a little extra to try it for myself!
If you think bacon on ice cream is weird enough to cancel an order, I can only imagine you've never worked a customer service job.
Sure, in the most extreme cases it would be obvious to the crew. But simply making mistakes at a higher rate than humans will result in a lot of unhappy customers.
Sure, but how do you distill this into a rule a computer can follow? "Suspicious" is not an objectively measurable thing that a program can just check against
Think the easiest way would be to collect order data for at least a good number of months if not a couple years and feed it in and use that as a baseline of what a typical human order looks like, anything that deviates too far from that baseline needs to be handled by a human until someone can validate it as a good order, though I imagine you could get false positives for new menu items unless you set a reasonable instruction for items that have never appeared in the dataset before.
The same applies for the ice cream with bacon on it
Have you never seen what Americans eat? Bacon Creaminators are excellent.
There are machine learning algorithms for anomaly detection though. They actually work decently well because exploits like this do in fact differ significantly from regular orders. Because they assume all anomalies are attempted exploits, their false negative rate is rather low while their false positive rate can be a bit higher.
Taco Bell has the capability to create a decently large training set from all recorded orders (which must all be valid and non-malicious) so they shouldn't have too many issues developing this model.
If an anomaly is detected, make a human verify it is indeed an irregular order.
This is handwaving, which, to be fair, describes a lot of AI "solutions". An anomaly could be as basic as a customer not wanting onions on their burger because the vast majority don't make that modification.
Now what do you do in that situation? Force orders to never have modifications? That customization is such an important feature to the point that burger king adopted it as a slogan with "have it your way".
The idea of anomaly detection is to project some input onto a (high dimensional), numeric output. From the training data alone, you can then see where the projections are clustered and develop a high dimensional "boundary" where everything within is known and good and everything outside is unknown and possibly bad. Since orders come in relatively slow, a human would be able to check for false positives and overwrite the computer decision.
By the way, an ideal training set is preprocessed and has duplicates removed and new orders added by recombining parts of individual orders.
For example, if we have 3 orders:
We could then create the following set:
And so on, and so forth. A naive variant is just taking the power set of all valid orders.
This is more complicated than just having the available menu items, the available modifications, and the limits on quantities to compare against. This is already available through the app/online ordering.
there is an incredibly finite number of ways to mess with this, they just need a button to send a report to the engineers with how they got messed with and eventually they'll have a complete list. I really doubt it'd take long to iron out the vast majority of ways that can be thought of.
A QA engineer walks into a bar and orders a beer.
She orders 2 beers.
She orders 0 beers.
She orders -1 beers.
She orders a lizard.
She orders a NULLPTR.
She tries to leave without paying.
Satisfied, she declares the bar ready for business. The first customer comes in an orders a beer. They finish their drink, and then ask where the bathroom is.
The bar explodes.
This isn't something you can input any text into, it's fixed, that joke doesn't apply, you can't do an sql injection here.
I don't know how you can think voice input is less versatile than text input, especially when a lot of voice input systems transform voice to text before processing. At least with text you get well-defined characters with a lot less variability.
No special characters, this is speech to text, inherently sanitized inputs.
Special characters is just one case to cover. If the user says they want "an elephant-sized drink" what does that mean to your system? At least that is relevant to size. Now imagine complete nonsense input like the joke you responded to ("-1 beers" or "a lizard"). SQL injection isn't the only risk with handling inputs. The person who ordered 18,000 waters didn't do a SQL injection attack.
none of those issues work because there is a whitelist of specific terms instead of a blacklist
-1 cannot be selected, a lizard isn't on the list of inputs, and my point with the sql is that this isn't a huge attack vector like an input field on a website, this is a dropdown list, essentially.
i challenge you to come up with one relevant attack that isn't order too much of a thing or order conflicting modifications (note of course the modifications are also essentially read from a dropdown list)
everyone here seems to believe that the input field paradigm is not solveable when the inputs are fixed, that isn't true.
Close one, a joke was related to but not a perfect match for the present situation. Something terrible could have happened like... Uh...
Let me get back to you on that.
Sounds like you've never programmed before.
If what I was saying was false, there would be the same issues with the touchscreen ones.
that's what happens 99% of the time. It's kinda been a trend on the anti clanker side of TikTok, just order a large amount of stuff so a human takes over and actually helps you
The point is that loopholes in software will always exist that lead to unexpected outcomes.
Why can't a trillion dollar AI say "Sir, that's not reasonable"?
It actually crashed, per the article. Presumably because it failed a sanity check.
You know what would reduce mistakes and speed up orders? Replacing the god-awful audio equipment that every drive thru seems to use. It wouldn't cost much to install good audio equipment and it would quickly pay for itself in time savings since people could hear each other more clearly.
Can’t find a source for this but I remember reading that the audio is intentionally bad because it subconsciously makes people more inclined/pressured to purchasing add-ons or upsells.
Hilarious, what a hero. Thank you king, this is the way to fight unsecure AI slop while they pretend its a legitimate reason to halve their workforce.
After hearing about this I now want to go up to AI menu board at my local Wendy's and order 17x10³¹ triples burgers.
Computer can't understand voices good enough so they think you say something different. For example standard voice to text has failed me for years. Voice control for years and years has still failed me. Using gps? fail every time because the computer or phone in this case thinks i say other words that sound nowhere the same. This is before and after AI. The biggest problem is voice recognition of words. Humans can separate sound and frequency better. Ear drums are better than microphones. Even in quite environments i have a voice that doesn't get translated properly and fails more often than it doesn't. So AI is going to glitch on a lot of voices too because the voice recognition part still works the same flawed way with flawed microphone technology.
Ear drums are better than microphones
What do you mean about this part? A quality microphone is basically the same thing, although the impedance matching is achieved different ways.
At least i want to believe ear drums being flesh vibrate differently than non organic materials, cells squeeze and contract. The shape of the ear drum is different and the electric signal in the human body is different than electronic devices. It's just a tiny tiny amount in a human. higher voltages like outside the body change and shape the sound a bit or cause distortions of certain wavelengths of the sound. Just my general idea. I had an ex girlfriend who had a wind up shortwave radio and if using it wound up instead of plugged in to AC or battery she could hold a signal on to a weak station. And no matter the power of the other stations there was less static and distortions on easy to get broadcast or hard to get broadcast. This was just a pocket radio. The flow of electricity is different between those method ac vs dc. Was the kind that winded down as you use it. Power is more stable during wind up or something. I could be a little bit wrong on the radio part because i would have to have my ex girlfriend show me again and no way am i going to go talk to her after over 15 years.
One day in 2012 at McDonald's drive thru, talking to a real human on site through the speaker just like the good old days, I ordered
"A Filet o' Fish."
Imagine my surprise at the dollar total and when they handed me a bag of
EIGHT Filet o' Fish.
I can’t imagine what it is reliable enough for.
Apparently it's good with knots.
it's not 'ai', it's just a poorly trained voice recognition system that's trying to decipher any random person's voice. voice rec has a difficult-enough time when trained on a single person in a professional setting (lawyer, doctor, etc)--which is work we used to do here. it's going to absolutely fail in a 'drive thru' environment.
it's not 'ai', it's just a poorly trained voice recognition system that's trying to decipher any random person's voice.
I’m baffled that you can say “It’s not ‘AI,’ it’s a machine learning powered speech to text system” with a straight face.
Even if we were to agree that ML-powered speech to text isn’t AI (and I don’t agree to that premise, for the record), there’s still the matter of processing the transcription to transform it into something that can be understood by the point of sale system - aka natural language processing. And while that NLP could be implemented without use of an LLM, given LLM’s current level of hype and the ease with which they can be shoved into any given product, I wouldn’t bet on Taco Bell execs approving such an approach, much less asking for it.
And none of that is AI, if we use the term in a way that's useful and stop speaking in salesspeak. Machine learning algorythms, LLMs, all of that isn't AI in any meaningful sense of the word. It's all semantics of course, but the more sam altmans of the world are trying to blurry the line between intelligence and an algorythm that can identify a picture of hotdog, the more we kinda need to be stickers for the proper terminology. It was kinda cute at the beginning of this crazy cycle, but now when megacorpos are demanding all your data and putting their half-baked chatbots into everything, it's not cute anymore.
Please, enlighten me - how do you propose we use the term “AI” in a way that’s more useful than a definition that includes machine learning, large language models, and computer vision?
I doubt I’ll agree with your definition, but I’m curious to see how you would exclude machine learning, computer vision, LLMs, etc., from your definition. My assumption is that your definition is going to be either a derivative of “AI is anything computers can’t do yet” or based on pop culture / sci fi, but maybe you’ll surprise me.
To be clear, I’m a software engineer; I’m not speaking in sales speak. I’ve derived my understanding of the term from a combination of its historical context and how it’s used in both professional and academic contexts, not from marketing propaganda or from sci fi and pop culture. I’m certainly aware of the hype machine that’s ongoing, but there are also tons of fascinating advancements happening on a regular basis, and the term “AI” is at minimum a useful term to refer to technologies that leverage similar techniques.
"It's not AI, it's just <multiple things that all fall under the category of Artificial Intelligence>".
AI is a huge field of computer science. It's not the one tiny narrow definition of artificial general intelligence like HAL 9000 or Skynet or Detroit Become Human.
all 86 comments