Most barcode scanners are already cameras, actually. Turns out that it is a lot cheaper to manufacture than the old "put a mirror on a moving 2d servo assembly" approach - and it's more reliable as well.

You can get an entry-level barcode scanner from a reputable brand with 2d support for less than 100 bucks, and you'll have trouble finding any kind of premium scanner which doesn't have 2d support.

And cheap generic one for about fourth or third of that... With some software bugs, but still surprisingly capable. Like reading kilobyte sized QR codes.

Got cheap one from China that does my uses cases pretty well. Supports USB, proprietary wireless and BT...

There is cheaper models with less features. Pointing towards quite a lot of market segmentation. And I could guess some of those models are same but with different firmware...

If camera’s are cheaper then why do supermarkets stick to lasers?

I’ve worked in areas where barcodes and QR codes were used for automation — we used both but for fast, cheap and reliable scanning barcodes were the winner

Where more data needed to be encoded then QR codes were used but the wuality and price of cameras needed to read QR codes as items came based on an automated line was significantly more costly than barcodes

Can any regular smartphone scan those RFID tags? If not, then the QR codes are superior because they can be scanned, making it trivial to use them for product recalls.

People have been making the claim for at least 20 years, and it still hasn't materialized yet. The problem is that it is a nontrivial increase in manufacturing cost, and you only get a benefit out of it when every product in the store has adopted it. If you want per-product RFID tags, you first got to convince an entire industry to eat a significant cost increase for a few years.

Besides, it has gotten less attractive over time. With the rise of self-checkout stores no longer save on cashiers by switching to RFID. The kind of tag to support "scan an entire basket at once" isn't compatible with being scanned by a NFC-enabled smartphone, so you can't use them for extra recall information either.

Serialized printing is basically a solved problem by now. I have several dozen products using it in my kitchen pantry right now, and none of them suffer from smearing or have low resolutions - and I can guarantee you they were rapidly and automatically applied in mass produciton.

Yes, but it's no different from the QR codes already present on many consumer products. In fact, it is easier to spot, as either a cashier (or someone at a self-checkout) is likely to notice that there's a sticker over the barcode and get suspicious that there's some kind of fraud going on.

This is already the case. Virtually every single food product comes with an expiration date label - which includes some kind tracking. You pretty much need to have that to do any kind of sensible recall when there's yet another poop lettuce outbreak. Looking at my pantry it usually seems to be include kind of lot number, time of production, and/or production line identifier.

I reckon they already have a pretty reasonable idea of which store it ends up going to, and with just-in-time resupply a pretty reasonable bound of when it will end up having been sold. Look at the overlap of a few dozen products and it should be quite easy to reduce the number of people who could've bought all of them to just you.

The QR codes just make this more explicit, and slightly easier to track. You're not wrong, but at least this time we could get, say, automated warnings in the Walmart app of a recall on something you bought out of it.

Today, with UPCs: We can reckon about whether or not the stuff we buy could be pinned down to a given store, or a transaction. There's lot codes and sell-by dates and stuff on the stuff in my pantry, too. But none of that was recorded when I purchased it.

So it's still a deductive process at best to correlate a lot number and a consumer. If all we have to go on is the implication of a limited range of lot codes, then the results are fuzzy.

Tomorrow, with unique serialization: The right person will absolutely be able to pin down exactly which store a particular can of beans was bought at, and when, and by whom. It's not just something that is made "slightly easier" -- it is instead a fundamental built-in capability of the concept. There's nothing to deduce or to guess at when this unique data is collected and correlated both deliberately, and automatically.

(RFID product tags, commonly known as EPCs, are also usually unique. This property reduces error from duplicate reads. When each tag is unique, it disambiguates a checkout involving 3 packs of #2 pencils from a checkout with just 1 pack of #2 pencils that was read 3 different times.

To pick one standard: SGTIN-96 is often used for EPCs on individual items and includes 38 bits for a serial number. That's enough bits for ~274 billion unique numbers. So with a haystack of 274 billion packages of #2 pencils sold and scattered around the world, the one in my desk drawer is very easy to identify.)

Because the consumer has to be able to trivially scan it with their smartphone and get useful information out of it. In reality, that means it has to be a valid URL you can visit with a web browser, as forcing literally everyone to install a proprietary URN-handling app is a dealbreaker.

The whole point of it is that you can just say "scan the barcode with your smartphone to find out if your product is part of this recall".

URN that can be copied and put in any national recall website seems better to me. I don't understand who I am supposed to trust to keep domains with the right content on them instead of porn sites. Seems very high faith for clown town.

On-demand printing. It's how expiration dates are currently already being printed on products during production. The only difference is that it goes from "put it literally anywhere" to "put it inside the space intentionally left blank".

And that's exactly why they went with QR codes. The 2D barcode for checkout just contains a valid URL in a well-known format, so it can do double-duty as link-providing QR code for customer information. And because QR codes can store a lot of data, adding the batch information from the current tracking-only Datamatrix is trivial.

There will be no "wrong" barcodes to scan, just redundant ones for legacy devices.

> People don't do this because it leaves tons of money on the table.

In my country I literally got a letter from the government if I could please file my taxes, because they believe the automatic deductions are too high so I am likely owed a back payment. In a previous year (I am not very good with non-timing-critical paperwork) they even called me on my personal cell to inform me about something similar.

The slogan of our tax collection agency is "we can't make it more enjoyable, we can only make it easier". If you're a regular employee and your taxes - deductibles or not - are a hassle, then that's 100% a political choice.

> AI-generated images can be indistinguishable from ones created by humans.

The people who make basic posters and menus by prompting an AI are neither using the proper tools nor the proper prompts to reach that level of quality.

> I also doubt most people care if something was AI-generated

A lot of people hate AI, even if only because of what it represents - which is giant tech companies completely overhauling your day-to-day life for a few cents of profit.

> They only care about the end result, which for AI-generated content can be as good or better, faster, and cheaper than what many if not most professional humans can do

Exactly, and the AI-generated content is 99% obvious slop. Could a skilled professional have used AI to make something decent-looking? Yes, but that's not what is happening. What we are actually seeing is someone who has never used AI prompting it for "a poster" and grabbing the first output. It doesn't even have the quaint look of work created by an enthusiastic-but-untrained amateur, it just looks plain bad.

If all AI output looked amazing and nobody could ever tell - or simply didn't care - then we wouldn't be seeing articles about "AI fatigue", would we?