Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

amazon.com

Post reply on HN
Print view View original

371–380 of 768 posts

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#371

Post by hibikir »

Reason077 wrote: post #245
moolcool wrote: post #132

Using Amazon for shopping is terrible, borderline unusable in 2024. They're hard to compete with because they're giant and have an amazing logistics network, but it also seems like there's a big vacuum in the market for an "everything store" that's actually good.

The worst thing for me is how the Amazon search algorithm seems to want to show you everything but the item you searched for. In many categories, even when explicitly searching for brand and model names, you’ll get dozens of off-brand substitutions and even random unrelated products appearing above it in the search results. Occasionally I’ve even noticed products that are available for sale (if you click on a direct…

In 2022, Amazon had 38 billion dollars in ad revenue. That's ads in that search page. Between the ad revenue, and variations in what sale is more profitable for Amazon, you get a lot of incentive misalignment. The page that makes Amazon the most money is not the one where the item you were thinking about is the first thing on the page. Giving you a worse page is just far more profitable.

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#372

Post by packetlost »

PheonixPharts wrote: post #118
packetlost wrote: post #27

I get that this is basically fraud and spam, but this should really highlight the dangers of letting an unattended LLM do anything for your company at all. It can, and will, fuck up dramatically sooner or later.

I don't find this any different than seeing an exposed jinja template: "{{product_name}} is perfect people who work in {{customer_industry}}" or the typical recruiter "Dear {{candidate}} I read your profile carefully and think you'd be perfect for {{job_title}} because of your experience at {{random_co_from_resume}}" If anything, I think it's kind of cool that we're seeing LLMs actually used for something very practi…

I don't think LLMs are evil either, but I think the real risks are extremely underplayed. This is a mostly innocuous example, but there are a lot of people trying to get LLMs into more places where the just aren't ready for yet.

The difference between a template is that the behavior is generally deterministic. Even if someone fucks it up, it means it's (usually) trivial to fix.

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#375

Post by xp84 »

GCA10 wrote: post #348
Reason077 wrote: post #245

Earlier quoted context omitted.

The worst thing for me is how the Amazon search algorithm seems to want to show you everything but the item you searched for. In many categories, even when explicitly searching for brand and model names, you’ll get dozens of off-brand substitutions and even random unrelated products appearing above it in the search results. Occasionally I’ve even noticed products that are available for sale (if you click on a direct…

Amazon's founding principle of "customer obsession" has been turned inside out -- at least when it comes to thinking of us consumers as the cherished customers. Those days are over. The new "customers" at the center of Amazon's business model are a global assortment of insta-merchants that don't make the products, don't handle their own logistics and don't have recognizable brands. So -- whoosh! -- in comes Amazon as…

And it’s bonkers how little they care about things that impact the customer. I’m a “Vine” reviewer (free* products in exchange for a review). Sellers game this system by listing a dozen or two duplicate SKUs and submit them to Vine, each in very small quantities (one non-skilled FTE could do the job of policing Vine for abuse like that, and they do not even care a bit to try.

*Note, they 1099 you for full retail value so really it’s just a discount of 100% minus your marginal fed and state income tax rate!

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#376

Post by wand3r »

joegahona wrote: post #286
Reason077 wrote: post #245

Earlier quoted context omitted.

The worst thing for me is how the Amazon search algorithm seems to want to show you everything but the item you searched for. In many categories, even when explicitly searching for brand and model names, you’ll get dozens of off-brand substitutions and even random unrelated products appearing above it in the search results. Occasionally I’ve even noticed products that are available for sale (if you click on a direct…

This is definitely the worst thing about Amazon. I pay them $120/year or whatever, and I search a specific product by a specific brand, and the entire browser screen shows me brands and even products I didn't even search for. I should not get ads in a store that I pay to use, especially in search.

I definitely agree that they don't return brand results. I don't use Amazon enough to remember the example(s) that were the final straw for me. Do you have any examples/remember which brand search you did? I am curious if some categories are "better" than others.

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#377

Post by bcrosby95 »

dan_bez wrote: post #250
jerf wrote: post #157

Earlier quoted context omitted.

To err is human. To fuck up a million times per second, you need a computer. Granted, here at the beginning of 2024, an LLM can not quite attain that fuck up velocity. But take heart! Many of the smartest people on Earth are working on solving that exact problem even as you read this.

"Fuck up velocity" goes straight into my vocabulary.

FPS? Fuckups per second?

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#378

Post by pants2 »

silisili wrote: post #149
moolcool wrote: post #132

Using Amazon for shopping is terrible, borderline unusable in 2024. They're hard to compete with because they're giant and have an amazing logistics network, but it also seems like there's a big vacuum in the market for an "everything store" that's actually good.

I'd really like a more 'curated' or vetted everything store. I don't need to see 40 of the same exact item rebranded into various English horrors. Walmart seems most primed to do this, barring third party sales. Or Sears, if they ever had a miraculous turnaround to their old days.

I'm prone to losing sunglasses, so some years ago I went through the process of testing out a dozen Alibaba sunglasses to find the best ones. I settled on one that's $4/pair, sturdy, and looks/feels/functions just like a $50 pair. Of course, being Alibaba, I had to buy it in bulk, so I now have sunglasses for life.

But that brings me to the type of site I want to see. Not curated luxury products like Le Creuset cookware at a markup, but curated dirt-cheap Alibaba products with low margins that have been tested and vetted extensively.

Massdrop or Monoprice are a little bit like this, but only for a few niches like headphones or cables.

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#379

Post by packetlost »

CobrastanJorji wrote: post #228
packetlost wrote: post #27

I get that this is basically fraud and spam, but this should really highlight the dangers of letting an unattended LLM do anything for your company at all. It can, and will, fuck up dramatically sooner or later.

Is this a dramatic fuckup? Because it quite possibly successfully created tens of thousands of listings more or less successfully. This one will probably generate no sales, but were there any consequences for this mistake?

The difference is the failure is non-deterministic and not predictable in any real capacity

Re: I'm sorry but I cannot fulfill this request it goes against OpenAI use policy

#380

Post by colinsane »

ketzo wrote: post #356
yunwal wrote: post #333

Earlier quoted context omitted.

I’m shocked that people are having good experiences with Amazon delivery in 2023. For me, 2-day delivery means it’ll get here in a week or 2. And forget about customer service that can actually solve my issue. I want whatever program you’re on

Where are you located? Not specifically, but urban, suburban, rural, remote? Anecodtally: I am in a super-urban location. I order Prime 2-day, on average, twice a week. It has been late maybe once in the past year; often, it comes a day early.

it can vary pretty tremendously within the same city. i lived 3 miles from the "downtown" part of Seattle in a house, and 50% of my Amazon deliveries were late by 1-3 days. my friends lived 2 miles further out from the downtown in the same direction, but in a 100-unit apartment and they never had issues.

not that 4 day delivery is bad. but promising to deliver something, and then regularly failing, is. i'd make plans for the thing being here by the promised day and just regularly be screwed.

Post reply on HN