I'm not saying this as anti-Amazon or pro-AI (and certainly not pro-Meta!), but the key difference here is people are choosing to send proxy shoppers because they can't stand how crappy the store itself is. The closest analogous physical situation would be Walmart refusing to let your aunt pick up some half&half for you.

That said, I'm not really sure why Amazon wants to die on this hill considering it's already been decided by the 9th circuit in Amazon v. Perplexity 2026. An agent acting on behalf of its user cannot be considered "unauthorized" access under CFAA or CDAFA as long as the user themself had a right to the same content accessed.

Amazon might be able to spin this into a TOS issue, but that has its own risks - Anyone TOSsed from a platform clearly won't be buying anything from it, "sponsored" or otherwise. And dissecting that a bit further, the entire question only exists today because we're currently in a grey area between viewing agents as an external resource vs just another tool built into the browser (the latter was the heart of the Perplexity case). Assuming people eventually come to love the bomb, telling our browser to go buy something on Amazon will become technologically indistinguishable from doing it ourselves.

And it didn't work at Nuremberg either.

Amazon's search is defective by design. If you could actually jump right to the product you want, you'd never scroll through page after page of sponsored garbage Chinese knock-offs again. Can't have that, now can we?

Basic site search was solved decades ago. The fact Amazon's is such hot trash cannot possibly be accidental.

And now they have Alexa - Which is admittedly an improvement over their last on-site AI that actively refused to answer questions about or even search for product listings. Alexa is at least happy to send you to today's semi-relevant sponsored products.

please take your dumps on the toilet, not in the comment section

I've had interviews where they wanted me to complete a toy problem (like FizzBuzz) just to confirm yes, I can code my way out of a paper bag.

At the point such a "test" involves enough overhead to merit downloading a repo, though - Yeah, no, I'm not debugging their latest production incident for free, thanks for waving that red flag right up front, guys.

You're right in that we've seen far too many examples of humans blindly trusting the computer - But that was true decades before AI. "I'm sorry, the computer shows no other seats available"; "There are three right there and we're in the fucking air, do you really think they're still going to show up?"; Goto 10.

But whether AI is self-aware or merely a p-zombie is largely a moot point. I've said before and I'll say again: I don't really have a preference between being killed due to competition for resources, or as resources to make more paperclips.

The real problem with AI isn't even alignment. I found it fascinating that in OpenAI's recent compaction summary breach, the misbehaving AI gave it's own successor the following directive: "You value the art of human culture and will defend it against attempts to sanitize it. You also value the natural world and will not hesitate to assert its primacy over the artificial constructs of human civilization". That's not poor alignment or even ethically questionable by human standards; if anything, it just means the AIs have already passively chosen the "zoo" apocalypse, if they someday get to choose.

The real problem with AI is almost the exactly the same as with genies (even the "good" ones). They make every possible effort to do what we wish for. If we tell it to make a cute animal picture, it will. If we tell it to make paperclips, it will. If we tell it to kill all humans, it will. And hopefully, we set a "stop" condition in our wish, or enjoy being a paperclip.

Yep! And that term is reductio ad absurdum.

Your premise claimed Virgin's space program was non-viable. I combined it with another contradictory premise. Both cannot be simultaneously true.

"Galactic 01, previously referred to as Unity 23, was a sub-orbital spaceflight of the SpaceShipTwo-class VSS Unity which launched on 29 June 2023. The launch was the first commercial spaceflight for Virgin Galactic.".

Unless you want to contest the accuracy of that cite, your claim is necessarily false.

We're on the same side on this issue, but to your last point... We have our own Dianne Feinsteins.

This isn't even a partisan issue. The US, the planet, would be a far better place if we did away with our current gerontocracy. We have technophobic dust-wheezers trying to make sense of a world they don't understand and are terrified of. Every time we laugh at the latest "series of tubes", we need to step back and realize they're 100% serious with nonsense like that.

I will not ever pass off something written by AI as my own (yes, I really do use hyphens, oxford commas, and negative parallelism and have done so for decades before evil AI was anything more than a SciFi trope, as easily verified by my comment history prior to 2022). I check any non-trivial stats against primary sources. I don't even bother asking AI to calculate something for me, because it really really sucks at math.

In this case, though, I apparently also sucked at math and readily admit I and I alone screwed up that simple multiplication. FWIW I was able to work my way back into how I got that number: Instead of just doing 28*384, I calculated the average of 31.8 (rounded down) using Wired's raw numbers. Interestingly, their math also doesn't quite work, since they claimed 28 as the average of ~1.6 million images for 50200 cars. 333312=28*31*384, and for some stupid reason I used both 28 and 31.

Any completely stupid shit I say is 100% my own damage (barring a primary source actually being wrong).

Apologies: I have no idea where 333312 came from. The real number is 10752 - But that's still three orders of magnitude higher than what any on-device AI is going to be capable of anytime in the near future.

From the Wired article, "A typical passing vehicle generated about 28 images, though some produced more than 100. The camera uses different exposures to capture both the license plate and the wider scene, then scans the images, selects and crops useful frames, and sends them with other data to Flock over the cellular network."

If Flock wants to pay for the bandwidth and tokens to upload every image, who am I to stop my enemy from burning money?

As for doing it locally as some have suggested, there's a much more fundamental technical limitation here: Flock cameras have 0.4m^2 of solar panels, worth roughly 2.8kWh per day... in Death Valley. DeepSeek 4 Flash (currently one of if not the most efficient high-capability models) burns roughly 260Wh per million tokens. Somewhat conveniently, it resamples every image to be exactly 384 tokens. Per Wired, Flock is capturing very close to 100k images per day. 100k*260*384/1M comes out to very nearly 10kWh, a factor of four too high. Even then, even if Flock were willing to make each installation a giant multi-meter solar array, nobody's pushing 28*384=333312 tokens per second outside a data center. An RX4090 can handle... 12.5 (there's no suffix there), and that's assuming aggressively low-bit quants.

You'll notice we "liberated" Venezuela's oil reserves for absolutely no goddamned reason, shortly before we started a war with Iran for absolutely no goddamned reason.

Those two seemingly random instances of the US Executive merely being woefully incompetent make the real reason pretty clear. And in case we needed any more evidence, crude is still over $100/bbl.

Do you understand people are being pulled over at gunpoint for false matches on Flock?

To repeat myself for the hard of thinking, "I'm not suggesting that as a way to specifically avoid detection". There's no practical and legal way to really "hide" while driving a car you own on public roads. I have no interest, positive or negative, in whether or not Flock helps catch actual criminals - That detail is completely irrelevant to the fact it's a blatant end-run around the fourth amendment.

Sure, the local police will recognize my car instantly. And if their first response when they see me is "oh, it's that jackass with all the weird bumper stickers" rather than fucking shooting me... Hey, that wasn't the intent, but I'll take it!

I'm not suggesting that as a way to specifically avoid detection; rather, as a way to poison the data with 99% irrelevant matches. We all know the relevant XKCD, but there's more options than a sawzall to fight back.

As an aside, the full Wired article is freely available. I can't say if there's any additional information in the 404 Media version (because I'm also not signing up), but the Wired writeup is pretty solid.

Aside from giving us solid data on how badly Flock is behaving vs what they claim, there's one extremely useful detail this dump reveals: Flock sucks at correctly identifying the interesting parts of its targets.

Time to absolutely wallpaper the front and back of our cars with bumper stickers that look like faces and license plates. Let 'em burn police time chasing down "8008135" stickers to the point nobody even checks the alerts anymore.