There is a lot of information in the article but little to draw a conclusion from. They bring up examples where the AI failed, but that alone does not indicate it doesn't work.
An AI that fails to recognize some thefts but does recognize others can still be considered to work as long as enough thefts are caught. Imperfect, but given the low cost of a failure relative to not using an AI, imperfection seems quite tolerable. Even the false positives seem to be of little concern as they involve a shopping associate coming over and having to assist an individual. As long as the experience for the shopper remains more pleasant than waiting in line for one of the cashier operated checkouts, it is still an overall win.
So yes, the AI is imperfect, but where is the information that justifies the conclusion it doesn't work and not the conclusion that it doesn't work perfectly?
I think what people fail to recognize is that AI/ML solutions will _always_ be imperfect.
But, they scale. There's always a threshold for the accuracy/efficiency point. It's not meaningful to identify errors, even if they seem glaring, if overall you're reducing 80% of loss at 10% of the expense.
Agreed. Which is actually an upside to an automated, but imperfect solution as reinforcement.
I've just been generally surprised at how often I see an ML solution implemented the complaint is "it isn't perfect", when perfection was not the goal metric. I think people just expect robots to do everything exactly right.
They must have forgotten the name of the game is loss prevention for the business and not about stopping shoplifting. Cutting payroll on your loss prevention is loss prevention in of itself.
I’m not sure. When self-checkouts were introduced here, they had staff supervising to help and educate customers then significantly scaled back that headcount.
Fast forward several years, there’s more staff stationed with self-checkout due to theft.
I think retailers are going to calculate the cost of headcount and theft and make a decision based on an equilibrium.
Of course, the best outcome for Walmart is to have an automated LP system to reduce staffing cost for the long term.
I used to regularly watch staff at a grocery store i used to go to not give even one ounce of fucks as people would walk through the self checkouts with buggies full of groceries, not even pretend to go through and walk straight out of the store. Even the security guards would just shrug and not do anything.
That store must have lost thousands every day from that, but they're an insanely busy store and probably more than make up for it in sales.
Also, I remember when I worked at a grocery store, the policy was if you seen shoplifters you weren't supposed to approach them or do anything other than to just tell the manager. There were a few thefts from that place and this was before the days of self checkouts. I remember I seen a guy once stuff some things in his backpack, I told the assistant manager, he followed him around through the store, but in the end the guy just left and nothing really happened. He kept whatever he took.
If they’re open 24/7/365, that’s still six figures in savings. And if one of those employees is watching the self checkouts then losses might not even be that much higher. And that with 90s technology — very little/no CV. I think Cashier as a stand-alone job will probably die off in our lifetime. Maybe you still have a store manager who is up at the front of the store whenever they are not doing something else.
One time I had $200 in self checkout and the machine broke and no staff were around so I gave up and walked out. Store lost the sale and whatever goods were considered ruined and I won't go back to that store. but they also saved on staffing costs.
It may very well be true, and in fact reasonable that there's a lot of false positives and that 'it was designed that way' due to obviously an inherently higher error rate.
So it's really a that the parameters of the tech are overlapping poorly with the parameters of customer convenience, perception, treatment, and other aspects of the next consumer experience and op. costs.
This is probably a great example of understanding the limitations and special needs of AI in the real world. It's not magic, it's just 'some magic' that we have to develop processes to work with.
It is a bit strange how the in-house and other competitor is also mentioned, Walmart are big on data warehousing and management, in a lot of cases they are pushing innovation and r&d, in some cases for decades. This article needs to show a more detailed analysis of when it does actually work, a fraud KPI ?, or something else besides some anonymous complaints.
You can trick any system, but that's not telling on it's overall effectiveness.
Their new kiosks are so infernally slow that I like to imagine they're electron apps running on a raspberry pi, which would make them the very peak of innovation.
Same for McDonalds' ordering kiosks. They feel webtechy in the worst way. Plus they're horribly organized, probably to try to boost sales. Unless the line at the counter's over 4 deep I can't imagine the kiosks being faster, even if there's no wait for them.
Tech has been Wal-Mart’s competitive edge for decades. They had an IBM and electronic POS since 75 for inventory management. They had a company wide network in 77. In 87, they installed the largest satellite network at the time. Allowed two-way data (inventory and sales) and voice communication and one way video to all of their properties. They built an enterprise DW in 90 and put EDI through the internet in the mid-90s. There is a reason they could beat every other retailer on cost for decades until Amazon came around.
Sorry but a citation is not needed, they have poured vast amounts of money into IT infrastructure, logistics, supply/demand, warehousing, POS, RFID, the list goes on. How do you think they became so large? Technology you take for granted everyday started out funded and developed by Walmart (usually in collab with IBM, TI, etc). They have brought money and the desire to the table in tech for over 40 years, often pushing the envelope of what's even possible.
If you have kids it's really easy. You don't notice that they put an item in their stroller or inside another item that you're buying (like you're buying a cooler and they put a toy in the cooler).
Technically it's your kids stealing, but since you are responsible for their actions, it's on you.
When this happens to me I usually let them know the next time I'm at that store (or if I discover it in the parking lot I go back in). Sometimes they charge me for it, usually they just say thanks for being honest.
Technically you can't, as being inadvertent means you lack the mens rea to commit theft. More colloquially it is used to describe any accidents in your favor with terms of paying for purchases. Forgetting to scan an item, accidentally having two items get stuck together and only scanning one, and similar incidents.
I got home to find a package of microfiber cleaning cloths in one of my bags once. I have absolutely no idea how they got there. But I guess I 'stole' them. Was going to bring them back to the store but realized that would probably be super weird so I just kept them.
I've also walked out of Lowes with a couple dollars in random hardware in my pocket before. Went in for some screws so didn't grab a cart, decided to pick up some other stuff since I was there, and stuffed them in my pocket temporarily. Forgot about them during checkout. I actually went back and paid for those though because I felt like such an idiot about it.
They have that hand scanner tool, so you can scan larger objects without picking them up, and I often will forget I have smaller objects under the larger one I scanned.
Also, scanning 1 item while holding two of it in your hand.
I've had occasions where the self checkout has had an issue after scanning something and the staff come over and login and do something and you carry on. It might only be when you check the receipt later at home that you realise that those carrots or whatever that caused problems were just voided off the bill/never scanned at all but were allowed to be bagged so actually you didn't pay for them.
Often the staff look like they are on autopilot and aren't reading anything and just going through their shift on muscle memory so I imagine this happens a lot. Is it stealing if you tried to pay but the staff did something on the machine and it didn't charge you? I don't know.
I've done it in the UK (three times) by failing to notice my contactless debit card has not been accepted and walking off with the newspaper/sweets etc. (I've gone back later to give them the money, but this is seems surprisingly difficult for a large store's systems to cope with.)
This is a very hard problem and thinking a product could do this without extremely high error rates is naive.
Any implementation should use this as a supplement only, not replacement for humans. For example a human watching 10 checkouts and quickly manually reviewing scenes flagged as suspicious. Any implementation should also account for the fact that loss will still occur and account for that in the RoI.
The fact that the technology is very error prone is not surprising (to anyone that actually builds AI tech). The issue here sounds like poor implementation and unrealistic expectations from management on what AI truly can and can’t do.
Preventing scanning one item and placing two on the bag? It's an easy problem. Just have scanned items be placed on a scale. People will be bagging stuff to take home, so just have bags placed on a scale and observe the weight increase as items are placed in the bag.
You probably know the weight of one jug of milk, but even if you don't, it's easy to "learn" the weights of new SKUs statistically.
Early self-check kiosks did this very poorly. It was super frustrating as a customer to be constantly asked to "please place the item in the bagging area", even when I already had, not to mention the hassle it caused when buying something you don't want to bag. Eventually I'm assuming they gave up on that and turned it off.
Checkouts in the UK still do this. They just seem to have gotten better at it, so I rarely have problems anymore. Possibly they loosened the tolerances.
Still pisses me off endlessly though, I'm only scanning a couple items that are going right in to my backpack. Takes 3x the time because of those stupid scales.
It seems to be still on based on the fact that every self checkout has a scale and Walmart does have bag item prompts until you bag it. It isn't as annoying though.
My guess is that they have more tolerance for items not of the exact weight and/or more accurate scales.
I use the self-checkouts pretty much every time I go to walmart. I frequently do not put items on the scale after scanning them, either because they're too large to bag or because I've run out of room. I assumed the scale wasn't running.
This has already been tried. I, personally, hated it so much that I just stopped using the self-checkouts whenever possible, because employee checkout was faster, even with the wait for customers queued before me. The weight check always brought the checkout process to a screeching halt, requiring an employee to come along and restart it, even with the smallest baskets in the best of conditions.
Is this not standard everywhere self checkouts are used? All of the ones I've ever seen used scales and enforced the weight matching the scanned item. Of course, there are still possible cheats: not putting the item in the bagging area, for example, or telling the kiosk you are buying carrots and then weighing something more expensive.
I am thinking of the opening scene in "Raiders of the Lost Ark", where Indiana Jones is trying to replace the golden idol with a similarly-weighted bag of sand.
I did that weighing in a store and it was an absolute disaster. Cumbersome and error prone, especially when you have larger items etc. I think that has huge hidden losses in people choosing manual checkouts or worse - avoiding the store all together.
Self checkout without that works well in many supermarkets, IKEA etc. There are losses but with random checks they are lower than the staff costs - so It's a win.
I'm suprised they even bother to try and stop it. My grandfather works at WalMart, and said that they're told to not try and stop shoplifting because they don't want customers to be afraid to shop there.
Some around here hassle people for their receipts near the exit. It was already my least-favorite place to shop for a ton of reasons, so that was enough to stop me ever going. It's one thing at Costco where I agreed to it and they're very fast and nice. I'm not standing in line to get permission to leave friggin' Wal-Mart (yes I know you can just ignore them but they sometimes try to stop you and make a big deal of it and that's just not a situation I want in my day)
Also Best Buy, which is particularly annoying as they watch you check out at the register five or ten feet from their post, then "randomly" (almost always) demand the receipt.
Who are you checking, the cashier or the buyer? My guess is a deterrent for some kind of collusion.
I believe this varies a little state-to-state and even in those (which I think is most?) where it's true they can usually still legally detain you if the have reason to believe you're trying to shoplift, though refusal to show a receipt on its own is, obviously, not enough to pass that test. Also workers are unevenly educated on the topic so may try to get in your way or otherwise escalate things if you do this, so take care and stay cool.
They do have the authority to prevent you from returning to the store. This is the only reason I ever agreed at Fry’s.
At Best Buy I stopped shopping there for years. I don’t remember why I eventually went in one again, but in the years since I’ve never been asked once. I do wish they’d get rid of the guy at the door entirely. It’s weird. And they always greet you from the side or back.
Are you sure? Receipt checking cannot be forced. I'm surprised corporate would allow that, since they know they can't force the issue so they cannot stop shoplifters.
Your local Walmart has your face on camera dozens of times before you get to checkout. If you don't want Walmart storing images of your face, shop elsewhere.
If it bothers you so much, why not just use the regular checkout? Or just shop somewhere else? Do you cover their security cameras all around the store with bags or just the self-checkout one?
This seems a really odd place to draw the line on surveillance.
It just seems odd that they've captured your face walking in and all over the store with their other cameras, but then you choose to cover the camera at the checkout. Why? What do you hope to gain/prevent by doing that? Why are you risking getting in trouble?
IMHO this is smart. They're training ML models using this data (and it feels intrusive to me) and one might assume this is higher quality than the cameras scattered about the store at odd angles.
While not super related, I've always been fascinated by shoplifter analysis of loss prevention you can find on the internet: https://raddle.me/wiki/shopliftingml
Is it just me or do the "Concerned Home Office Associates" not sound like they actually work at Walmart? I talk to the workers at my local Walmart and they don't refer to themselves as "associates", or care enough about a minimum wage job to contact media over anti-shoplifting AI failing.
I bet they do. "Home Office Associate" means office worker or manager. A programmer falls under this. It isn't a minimum wage "store associate". Of course they won't refer to themselves as store associate when casually chatting to a shopper.
And that is why such technologies are best deployed with humans in the loop. Tech flags, human checks, human label is fed back into training. Fewer humans will be needed now, even fewer 2-3 years from now. There are literally tens (if not hundreds) of thousands of people every day doing nothing but watching security feeds. And 99% of the time nothing is going on.
Walmart self-checkout is the second best working self checkout I've used so far. (The best one is Sam's Club: you scan things with your phone as you put them in the cart, and when you're done you tap checkout and leave the store. Payment happens in the phone app.)
If the cost of having self-checkout machines that actually _work_ and let you complete your purchase is that some items slip by, that might well be worth it.
In contrast, Safeway chain self-checkouts are such a miserable experience that for more than two items, even waiting in a long cashier line is worth it. Their system requires you to place every purchased item on a scale after the scan, and a good fraction of the time the scale fails to read the expected added weight, forcing you to wait for staff to unlock the machine again. For a cart full of purchases, this will happen half a dozen times.
I mean, presumably Wal-Mart has excellent statistics on "shrinkage" from regular inventory checks like any store, as well as recording how often shoplifters are caught, and so they have clear data as to whether or not Everseen is an improvement or not.
And I don't understand what this has to do with COVID-19 at all in the article.
Finally, what do employees care about theft? It's not coming out of their paycheck. False positives from an anti-theft system seems like a small job nuisance at best -- not something to whistleblow with a "slickly produced video".
Nothing about this article makes sense to me, unless it's motivated by an Everseen competitor -- whether another company or an internal division of Wal-Mart that thinks they have a better solution. In either case, why would the general public care at all?
> Finally, what do employees care about theft? It's not coming out of their paycheck. False positives from an anti-theft system seems like a small job nuisance at best -- not something to whistleblow with a "slickly produced video".
I think their is a massive divide from Walmart's Corporate/Tech branch and that of the typical store employee. I overheard one employee saying back in December as we were days away from NYE that she had to remain at work despite having been approved for leave weeks earlier as most of her coworkers simply just didn't show up to work and demanded a raise from her manager. I now always wear earbuds when I go inside of one.
Whereas for the tech employee or risk management office drones they don't face those concerns and instead have to justify a growing budget for their department, perhaps even their jobs in this environment, with these kind of antics.
I inadvertently noticed how easy it is to use their buggy and slow check-out kiosks against them, but since I only ever buy motor oil at Walmart anymore I don't see this being a useful risk. I actually went back and paid for the item when I noticed it made the noise that it scanned but didn't include the item to my checkout price, which was a total pain in the ass that meant standing in line at the nearly always understaffed customer service/return area.
Bobby Lee, the ex-owner of massive Bitcoin exchange in China and the brother of Charlie Lee creator of Litecoin, was ex-Walmart tech. In a recent podcast outlining his previous history in the Bitcoin ecosystem he made it clear they have no shortage of talent there. To what end, is another matter entirely.
When I worked retail, employee theft was the big deal. From pocketing cash to a dock guy bringing up stuff for his buddies during busy times to the Theft Prevention team doing something to all get fired I think the employees did more damage than the shoplifters.
self check out systems are great and the concerns of the workers should be heard but it comes down to not having all the numbers.
what is the positive to false positive rate? projects can have metrics that those in the field are not privy too and it could down to they are preventing a sufficient dollar amount to not consider the false positives are an issue. Again, what are the percentages?
See also farmers' produce stands that are often completely unmanned and run 100% on the honor system.
People take fruit and vegetables, and put the money in the box--or if the farmer is tech-savvy enough, through their phones. And even if a few people don't pay, they still get more money from their inventory than shipping it to a wholesaler or a grocery store. (As far as I know, direct to a local gourmet restaurant pays best.)
For an extra employee on the duty roster, you're probably paying minimum $7 per hour, every hour that store is open to the public--probably 16 hours a day, every day, for a big box. That's $41k per year, plus the per-employee costs for 2 or 3 people, which are probably another $60k, unless you're an employer that doesn't offer benefits (i.e. Wal-Mart).
The employee would likely have to stop $200/day of petty theft. That doesn't seem impossible, but the problem is that if you pay your employee minimum wage, they don't actually care that people are stealing from their employer, because they have no actual incentive to do anything beyond what is the bare minimum requirement to avoid getting fired, which may be just catching one shoplifter per day.
And then, after all, if one caught every shoplifter, every day, people would stop trying, and one might miss quota on later days. The person whose job depends on shoplifters has a perverse incentive to encourage shoplifting, just as cobra-catchers have an incentive to breed and release snakes.
So the problem is more that it's too expensive to hire someone at a rate high enough to make them care about doing their job well, instead of just keeping it indefinitely while gaming the right metrics.
> People take fruit and vegetables, and put the money in the box--or if the farmer is tech-savvy enough, through their phones. And even if a few people don't pay, they still get more money from their inventory than shipping it to a wholesaler or a grocery store. (As far as I know, direct to a local gourmet restaurant pays best.)
It does.
> The employee would likely have to stop $200/day of petty theft. That doesn't seem impossible, but the problem is that if you pay your employee minimum wage, they don't actually care that people are stealing from their employer, because they have no actual incentive to do anything beyond what is the bare minimum requirement to avoid getting fired, which may be just catching one shoplifter per day.
> And then, after all, if one caught every shoplifter, every day, people would stop trying, and one might miss quota on later days. The person whose job depends on shoplifters has a perverse incentive to encourage shoplifting, just as cobra-catchers have an incentive to breed and release snakes.
You just explained the the broken window type fallacy of LEO and Police in general. When the crime no longer exists, you create it... by force. Oddly enough, it also applies to loss prevention.
> So the problem is more that it's too expensive to hire someone at a rate high enough to make them care about doing their job well, instead of just keeping it indefinitely while gaming the right metrics.
Agreed. And again, it applies to more than just shoplifting prevention.
1. British are ruling India
2. Concern about cobras
3. Establish dead cobra bounty
4. Cobra population crashes
5. Entrepreneurs breed cobras
6. Bounty program halted
7. Breeders release stocks
8. More cobras than before
The source story is just an anecdote, but there are plenty of historically verifiable examples of the effect, not necessarily involving cobras, specifically.
Yep. It may only take one or two false positives to lose a customer worth potentially orders of magnitude more than any particular item (depending on the price of the item, of course).
It should really allow you to continue scanning but prevent you from paying until the associate confirms that you're not stealing. That's 90% of the issue with the weight sensors -- the sensor stops me from moving forward with other purchases because it's making me wait for someone to check my stuff. But if they left the notification on while I continued to scan that would eliminate my issues with it entirely. Then they wouldn't need to rely on flaky AI.
The other issue I have is that the barcode scanners are not activated until the weight sensor settles out, which can take several seconds up to a minute, so it's way slower than going to a checker. If the weight sensor didn't gate barcode scanner activation I would have a much better experience.
All they really need to do is verify the weight of the total transaction and have the associate check on any discrepancies. And they can probably use a very noisy weight signal with a low-pass filter. They don't need to blow a ton of money on an AI solution that is about the same level of inconvenient.
If I had 10 or 20 million dollars I'd build a better self-checkout.
The self checkouts at my local Walmarts already do this. When an error occurs it flags for help but lets me proceed up until it is time to pay. By then usually an associate has already made their way over to check the errors in batch vs 1 at a time as they come up.
It's so much nicer of an experience compared to machines that just stop functioning until an employee can "help" you.
True, and the big ticket stuff at Walmart has additional protections that lower the need to risk a false positive on things like a candy bar or a shirt.
False positives leads to customer waiting and/or having to hire more staff to assist.
I know when the shops near me (UK) started rolling out automated checkouts I'd avoid the shops that had poorly calibrated systems that required frequent interventions because it was extremely annoying. False positives certainly had revenue implications for those shops.
So, I didn't read the article but if these employees didn't want to be replaced by API they could theoretically be helping the AI team find weaknesses in system and improve the system even more quickly.
Walmart are scumbags who will pursue shoplifting charges with no evidence and try to collect from innocent people. [1] I'm not surprised they are using AI for this, a system that's bound to cause false positives. It should be illegal for them to make such accusations and take up the court's and the defendant's time, money, resources, and freedom. False accusations should carry heavy penalties. Instead, they carry no penalties. Once again, another instance where the average person will get fucked over with no recourse.
An AI that fails to recognize some thefts but does recognize others can still be considered to work as long as enough thefts are caught. Imperfect, but given the low cost of a failure relative to not using an AI, imperfection seems quite tolerable. Even the false positives seem to be of little concern as they involve a shopping associate coming over and having to assist an individual. As long as the experience for the shopper remains more pleasant than waiting in line for one of the cashier operated checkouts, it is still an overall win.
So yes, the AI is imperfect, but where is the information that justifies the conclusion it doesn't work and not the conclusion that it doesn't work perfectly?