The problem is it’s Amazon Ai, and We all know how Intelligent that is, NOT
Let me share how “efficient” it is:
Booksellers know, despite AMAZON’s vehement denials, there are “ghost listings” that pop up occasionally. Maybe w/ other merchandise, too, I don’t know since our experience is books, only books. Deleted the darn title-or sold it-some poor buyer buys it months-or even years later! It’s a nightmare because one has to cancel, affecting SHIPPING PERFORMANCE.
This has happened to us so many times-we even have a template for it-which puts the blame squarely on AMAZON. TPTB no longer want us to use it-hey-they notify the cust. in case of cancellation. .
Due to a move, we’re culling-culling about half our inventory. .Being really, EXTREMELY careful. Deleting, deleting, going back and checking 30 minutes later to make certain the title is gone.
Recently, in the midst of this, we’ve had to put our seller account on VACATION. Once due to a graduation. twice due to illness/ death of a close family member. Twice, when we’ve re-opened the account, the first title to be sold is a "ghost listing: May have GUESSED IT-ONE OF THE CULLED LISTINGS.
When a seller account has been inactive for a week or two and, upon re-opening the account, is forced to cancel a first order, the PREFUFILLLMENT CANCEL RATE, which is supposed to be under 2.5%, goes through the roof. Ours has been in the stratosphere-55%, 65%! Dreaded suspension but other metrics were fine. Once we sold a few more, the cancel rate reduced over time -and most of ya’ll may have noticed that takes a week or so to shake off of the post-VACA order sluggishness.
THAT’s NOT THE WORST! On being forced to put the account on vacation again, the only titles reflected on our storefront are CULLED TITLES. Went back to delete-but can’t!! They’re not on our inventory, either active/inactive.
What you want to bet, when the time comes to re-open, we’ll sell one of 'em! Piss on Jassy’s “efficiency”.
8 Likes
Amazon AI is just the new term for… BOT
11 Likes
Not a books-only thing. Part of selling on Amazon.
7 Likes
Equally as efficient at misdiagnosing the problem and applying the wrong copy and paste response as any other undertrained entity!
6 Likes
I mean, lets face it, the forward seller facing employees are pretty lazy/inept/ignorant/stupid/mismanaged/imbecilic/etc., matched with policies and most likely software that are equally broken.
I like the idea that I can use AI to generate a prompt to request reimbursement, that will more likely be accepted by AI than J. Rakmapathi in a call center in Uzbekistan.
5 Likes
That’s one heck of an assumption.
3 Likes
My automated requests for FBA removal reimbursements have a higher percentage of acceptance than ones handled by lets look up the names of these “support” humans…
Sedukkarasi C., lupheng k., Arul Mozhi P. and Vishal G.
5 Likes
Once Amazon figures that out they will tune down the AI
8 Likes
I don’t handle FBA reimbursements specifically, because I don’t want to. But I am responsible for pretty much every other case or appeal, and my success rate goes up significantly once I get past the rejecto-bots to a real person, even if I can’t pronounce their name.
3 Likes
There’s no question that Amabots are typically poorly-parameterized - and always have been, since they were first rolled out, lo those many years ago - but to be fair, automated mechanisms aren’t inherently to blame for “Ghost/Phantom/Zombie” Offer-Listings.
The actual culprit responsible for this phenomenon is an intrinsic flaw in the original conception of Distributed Databases, which tends to manifest more frequently once those are scaled up for Enterprise Domain usage.
Due to that initial false assumption, the problem is @ best only mitigable, rather than eradicable - and such mitigation comes at a cost that Amazon, unlike financial institutions, government agencies, and the like - is unwilling to expend.
I’ve made no secret over the years of my disdain for AI, but I will admit that this is one area where I can see that a properly-designed corrective feedback loop could offer @ least some relief.
3 Likes
Amabots are AI, spell check is AI, Clippy is AI…
2 Likes
When we see “AI”, we don’t think of “ai” capitalized … we think of capital “A” and lower case “l” … as in “Al” Bundy
This dude must be all over the place.
4 Likes
The ghost listing problem is something I have expounded about at length, based on it being an insolvable problem associated with distributed database systems.
This was well known when I was still a technical person, one of the pioneers in computer to computer networking.
Much of the internet is based on assuming this insolvable problem does not need a solution.
Amazon was a pioneer in experience ghost listings because of the high volume of transactions it had. Ghost listings are becoming more common on smaller sites because they have reached enough volume for what was a relatively small probability of ghost listings is now leading to noticeable occurrences.
Relatives to layoffs. With or without AI, some layoffs at Amazon stand a high probability of providing improved services, fewer problems and lower cost. Real or artificial stupidity is bad for business. Read the Linkedin profiles for some Amazon managers if you wish to be enlightened.
4 Likes
Remember, AI had to cheat at chess to beat a computer. AI is not intelligent. It is predictive text. It is technically a predictive text large language model and has zero comprehension of math, reading, letters, or even logical deduction. Look up AI playing 20 questions, it is hysterical.
The choices companies are making to cut out the human, thinking factor will damage them in the long run. Never mind the damage it does to the environment between water and electrical consumption and mining for material components. But Amazon has never been one to care about any of these factors, to be fair.
This is not going to be a fun experience for sellers.
8 Likes
I’m looking forward to it.
I figure that Amazon’s seller service can’t get any worse, regardless of what changes Jassy makes.
But I do expect that AIs will not be as skilled at lying as many of the Amazon employees, so I will have a better chance of coping with them.
I see it as a net gain.
2 Likes
Prediction is the very substrate of anticipation, strategy, and meaning. Even a chess grandmaster is, functionally, a predictive engine shaped by pattern compression and bio-chemical recursion. There is a difference between the statements: It doesn’t understand like me v. It doesn’t understand. Damage to the environment as opposed to? 2m sellers selling carbon footprint free widgets? I think not.
1 Like
I just want to know when I can plug myself in, because the longer I am on this planet, the more I realize how many garbage humans exist in society.
Today I saw a blind “service dog” sniffing cheese in the refrigerated section of the grocery store. (Yes, the dog was blind, not the owner. Cataract’s like marbles). I also saw a man who only chained his tractor to his trailer with one chain, and damn near killed himself and others on main road through town when it came off the side at the intersection he turned at.
After I got out and checked to make sure nobody was hurt, I walked by the dim witted man and told him “you can’t park that there, please don’t breed” and drove off.
3 Likes
You should look it up. Predictive text is not the same as predictive strategizing, it only predicts words without understanding the meanings.
Look up playing 20 questions with AI it is hysterical and just repeats itself. No logical thinking or common sense, it only can predict text using internet data..or look up AI doing math, it fails constantly at basic addition when over a certain amount of digits. AI had to cheat to beat a computer at chess (there are full videos showing it).
And yes, one AI training leaves more of a carbon footprint than around 20 lifetimes of one American. But with 50B invested by large businesses, no one is backing down. This doesn’t even include drifting and the way AI degrades with more usage.
4 Likes
Quick responses:
You’re conflating syntax prediction with the absence of semantics. All cognition, even biological, is prediction. Ours is just bio-weighted. The idea that comprehension requires carbon isn’t falsifiable, it’s a metaphysical claim, not an empirical one.
Math? Neither could Shakespeare. You don’t blame a language poet for not being a calculator. GPT simulates the language around computation; it’s not built for math, it’s built for meaning modeling.
Carbon footprint? Apples to oranges. The real tradeoff isn’t AI vs the environment - it’s AI vs 10,000 redundant middlemen still using spreadsheets to drive inefficient global operations.
Degradation and drift? It doesn’t degrade. Context threads get saturated. That’s a constraint of current GPT deployments, not of the architecture itself. Sam’s personal GPT instance doesn’t face this.
Einstein also failed (failed entrance exam) - until he didn’t.
2 Likes