"For now, the in-app feature only surfaces AI images of clothing and home goods, allowing you to tap on the image that best matches what you’re looking for and search for similar-looking items.
In a blog post, Amazon positions the feature as a way to help you search for items if you can’t remember the name of a specific texture or style, like describing a “shirt with a draped collar” if you can’t think of “cowl neck.”
This is an important part. It’s not a failed idea that tries to make you buy non-existent products that are AI generated, they try to help you by showing you what you may think of and select the idea so it can help with the actual results.
You name a fair point here. I think the part where they’re using an LLM for natural language processing makes a lot of sense. Being able to describe something you don’t know the name of is a genuinely helpful feature. But you’re right that a better implementation would drop the wasteful image generation in place of searching up real images from their product library (which they’re still doing anyway because at some point they have to find a real product to sell you). It feels like that step was maybe added to make it “more AI”, probably at a manager’s insistence.
Oh almost certainly the point is just to have used AI. Theyre accelerating into popping the bubble. The tokens are starting to get priced more realisticly and there’s no way folks built things that financially make sense long term. Everyone is just being pushed to use it more to make more slop.
They can charge for the placement, or prioritize sponsored listing like they do for search results. So theyd make money on the feature, instead of spending money.
It should be fairly straight forward to check the state of a listing and ensure it hasn’t changed since it was evaluated for this purpose.
The way it (most likely) works is that you give your inaccurate a prompt to an image generator, “shirt with draped collar”, and it generates you a bunch of images of shirts with weird collars. You then select the one that looks kinda like what you want, and that is fed to an image recognition AI: “find me product pictures that look like this shirt with this kind of a collar”.
You can’t skip the AI because the whole point of it is that you don’t know what the thing you are looking for is called, so you can’t search for it. The no-ai version is the current search bar, you type in “shirt with collar” and start browsing items until you find it yourself.
"For now, the in-app feature only surfaces AI images of clothing and home goods, allowing you to tap on the image that best matches what you’re looking for and search for similar-looking items.
In a blog post, Amazon positions the feature as a way to help you search for items if you can’t remember the name of a specific texture or style, like describing a “shirt with a draped collar” if you can’t think of “cowl neck.” This is an important part. It’s not a failed idea that tries to make you buy non-existent products that are AI generated, they try to help you by showing you what you may think of and select the idea so it can help with the actual results.
How is this better than accomplishing the same general suggestion, but with real products? How could this possibly be cost effective?
You name a fair point here. I think the part where they’re using an LLM for natural language processing makes a lot of sense. Being able to describe something you don’t know the name of is a genuinely helpful feature. But you’re right that a better implementation would drop the wasteful image generation in place of searching up real images from their product library (which they’re still doing anyway because at some point they have to find a real product to sell you). It feels like that step was maybe added to make it “more AI”, probably at a manager’s insistence.
Oh almost certainly the point is just to have used AI. Theyre accelerating into popping the bubble. The tokens are starting to get priced more realisticly and there’s no way folks built things that financially make sense long term. Everyone is just being pushed to use it more to make more slop.
Suddenly, tokenmaxxing will no longer be a metric for performance once the bills come in.
Avoids giving free ad space for specific products, and avoids the product being changed messing up the recommendtion
They can charge for the placement, or prioritize sponsored listing like they do for search results. So theyd make money on the feature, instead of spending money.
It should be fairly straight forward to check the state of a listing and ensure it hasn’t changed since it was evaluated for this purpose.
The way it (most likely) works is that you give your inaccurate a prompt to an image generator, “shirt with draped collar”, and it generates you a bunch of images of shirts with weird collars. You then select the one that looks kinda like what you want, and that is fed to an image recognition AI: “find me product pictures that look like this shirt with this kind of a collar”.
You can’t skip the AI because the whole point of it is that you don’t know what the thing you are looking for is called, so you can’t search for it. The no-ai version is the current search bar, you type in “shirt with collar” and start browsing items until you find it yourself.
More consumerism is never worth it burning the planet to the ground even faster with more data centers.
I never said i agree with this stupid usage just explained that its not pointless + failed, just unnecessary simply.
deleted by creator