On Amazon, about 1/4 of the product reviews are either from people who haven’t received the product yet ("**** - I’m really looking forward to trying this out!") or complaints about the delivery instead of the product ("* - it took 2 weeks to get here because they delivered it to my neighbor’s house"). Amazon puts a lot of effort into removing fake reviews. These types should be comparatively easy to train AI to recognize.
Youtube - their search filters haven’t been updated in at least 5 years (and the last update was a disimprovement to remove the ability to search by category) - yet YT could so easily apply categories to uploaded videos automatically by image recognition. They’re owned by Google, right? Use some of the parent company’s tech! Then people could search for gaming, news, babies, pets, vlogs, fashion, etc. like we did 10 years ago, and not get so much crap in the results when doing a search.
Although I qualify for a COVID-19 vaccine, I still haven’t been able to schedule it. My state’s online program makes me type in all my information, happily tells me I qualify, then gives me a map of where I can go.
I then have to pick a dot from the map, which takes me to a link for the provider. I then am asked for my information again. Only after I enter everything does it check to see if there is vaccine available for me and each time so far I’ve been told “NO”. I go back to the map, pick another provider, and start from scratch.
I’d say the Youtube mess is intentional, not the result of laziness. Making navigation difficult increases the time people spend on the site, all the while potentially going off on an tangent / down the rabbit hole, seeing way more advertising than they did if the site was efficiently searchable.
With me it’s something that kayaker alluded to. When you make a call to a customer service line to a business and have to navigate through the automated picket line of various menus asking/requiring you to enter information about yourself ( name, address, phone #, DOB, account # blah blah blah ) to verify that you are indeed you. You wait and finally get to what you wanted to do in the first place: talk to somebody.
The first thing they ask is for all the info you just gave them “in order to better serve you”
It’s been recent big news (around select circles) that scalpers are using bots to immediately buy up new high end computer components and resell them at big markups. Basically the second something hits the major retailers’ websites, the bots buy it within seconds before a human can complete the transaction. These bots were actually repurposed and were previously used to buy up designer purses and limited run sneakers & other collectible athletic wear so they’re not new.
This feels like the sort of thing you could block much more effectively than they’re doing but I assume the retailers put just enough effort into it to have a plausible “Hey, we tried” while actually not giving a fuck because a sale is a sale and whether it goes to a happy end-user customer or a guy posting it on eBay at a 350% markup is all the same to them.
Maybe I’m missing some aspect of this, but it seems to me that if someone can make a business buying a product from company X and immediately reselling it at a huge markup, then company X is pricing their product too low.
The trick with most machine learning is that someone’s got to train the thing. Meaning, that once you set up your conditions and what-not, you have to feed it a set of curated data that tells it what is and isn’t “good”. Then you’d turn it loose on an uncurated set and review its decisions and teach it where it went wrong.
Ideally you set it up so that when it makes a decision, it gets instant feedback on whether it was a good decision or not, but that’s not always possible.
I think it would be hard to tell if it’s a bot or not except for the timing, and really, what the heck do retailers care? They sold out of their graphics cards- none of their business or concern what the buyers do with them afterward.
None of that really needs machine learning though; neither the bots nor the countermeasures need it to be effective. I’d bet the bots are more along the lines of complicated scripts.
I’d be willing to bet we start seeing machine learning/AI used in more things as time goes on- right now, it’s used a lot in things like ad presentation, natural language processing, etc… I wouldn’t be surprised to see weather prediction start using it and become more accurate. Same with medical diagnoses. Both are the sort of thing with a huge body of data that could be used to train AIs to predict and diagnose. But the stumbling block I bet is that it takes time and money to train the AIs to do it correctly.
It seems to me that someone could make a fortune tricking these bots into buying a bunch of junk.
Just put some useless products up on your site for a few seconds, and if they don’t sell out, take the ad back down before humans that would actually complain can see it.
I definitely agree with you in general, but some people are irrational, and would hate paying twice what a product normally costs even if the only other option would be to pay 3 times what it would cost from a scalper. They’d hate the scalper more but can’t enact retribution because it’s a one time deal. If they’d bought from the original company at the “outrageous” 2x markup, they’d tell all their friends / social media what a crappy deal they got from Company X and maybe even do less business with them.
In the Amazon vein, how many “Product answers” begin with something like “I don’t know…”
Yeah, that is just so stupid it should be prevented from posting as an answer, and the person answering that way should be banned from purchasing from Amazon for 30 days, including having their prime video and music perks suspended. All doe by AI bots.
Agreed! I also think there should be a “This was not helpful” button and those who are downvoted should be banned from rating or answering questions for a while.
I used to think that about the “I don’t know” people, but to be fair, apparently sometimes after you buy a product, Amazon sends you an email when someone else asks a question about it. If someone gets an email out of the blue asking them a question about a product they’ve bought, it’s not unreasonable for them to think that they should answer it, even if the answer is “I don’t know”.
A lot of that is (or at least used to be) because Amazon would email people asking for them to review things or answer questions about things they hadn’t even bought - maybe they looked at it several times, maybe went as far as adding it to their cart, or clicked through it on their way to buying one of the suggested items on the bottom of the product page, also the review requests were sometimes of the form “Is the Broots Midderpuddle good for instical glamnering?” - and the mechanism to answer “How the hell should I know? - I didn’t buy it!” was built right into the email.
I have received those emails in the past. If you took the time to actually read them, you would see they were soliciting information to try to answer customer questions.
So the AI could have been smarter in 2 ways: Ask the question more articulately so people know not to answer with “I don’t know” and automatically weed out the “I don’t know” answers from the absolute morons.
Amusingly, this happened with a video card where the same number “6800” was being used on the newest card from AMD (6800 XT) and was already an obsolete card (GeForce 6800) from Nvidia’s product line. The day the new card went live, the old card was instantly and magically out of stock on Newegg, presumably from bots that were just set buy video cards with “6800” in the name.
eBay and businesses with market place sellers like Amazon, Walmart, Barnes & Nobel, and Sears could be programmed to allow a buyer to stipulate maximum price including shipping and/or select an independent maximum shipping cost, which buyers would appreciate.
I very much dislike saying I’m looking for a wiget that costs no more than $25 and being shown market place sellers who have said wiget for $24.99 plus $18 shipping.
However I especially don’t like seaching on Amazon, saying I want to see Prime eligible items for $10 or less and being shown all sorts of third-party items too, so maybe I hope for too much…