This is an overly negative (“ruined the internet…”) take on the interface. I just tried it. I get a nice summary answer that has an “AI generated” disclaimer, and immediately below are actual reviews mentioning my search terms.
If anything I’m a laggard in getting on the AI hype train, but this seems totally reasonable and actually pretty useful (eg the product I was checking has no less than 11k reviews, so AI giving me a summary answer is also welcome)
It also nicely describes the shape Jeff's head and recommended I used a glossmeter to analyze it's shininess - so that's an added benefit of yet another company throwing LLM's at something they really didn't need to.
Absolutely agree. Being able to analyze and summarize reviews like this will make reviews much more useful. Especially given star ratings have become nearly pointless over the last few decades. Once polished, I should be able to do cool things like filter out all the reviews from people angry about their delivery service, or change the star ratings to a 1-10 score based on what was actually said. And certainly summaries of large sets of reviews as you say, ideally filtered on criteria such as region.
One of my top shopper-friendly wish list features for Amazon would be reporting MSRP price (or typical retail price if MSRP isn't available) and computing not just the "discount" (when applicable) but also any "markup" (when applicable).
Unfortunately Amazon benefits from opportunistic resellers/scalpers, so we'll probably lever see this.
1. Any announcement of a price reduction shall indicate the prior price applied by the trader for a determined period of time prior to the application of the price reduction.
2. The prior price means the lowest price applied by the trader during a period of time not shorter than 30 days prior to the application of the price reduction.
3. Member States may provide for different rules for goods which are liable to deteriorate or expire rapidly.
4. Where the product has been on the market for less than 30 days, Member States may also provide for a shorter period of time than the period specified in paragraph 2.
5. Member States may provide that, when the price reduction is progressively increased, the prior price is the price without the price reduction before the first application of the price reduction.
Especially ironic because Amazon blocks out mention of comparison prices in human written reviews. If a review says "I bought this back when the price was $25", the published version replaces the amount with "[...]"
Just tried this myself and was able to replicate. They even have suggested follow up questions for your random queries that don’t relate at all to the product.
Watch out perplexity, Amazon chat has entered the arena
I was having a bad day and this just cracked me up. Thanks op. I wonder if all search will just point to AI in the coming years, even within non-web applications, like MS Word (I know it is cloud connected but I generally don’t think of it as a web application.. I probably should).
You can install the Advigator Chrome Extension and click on the “customer review” tab on the top-right nav bar. You will get actual summary from reviews.
Because it's not correct, a search in a store is supposed to result in information about the products in that store, not unrelated garbage. Garbage is a problem in the internet, much easier ways to generate/access it in more places is an big increase of the scale of the problem, hence "ruin"
The input field at the top of bing.com says "ask me anything". Would you also call that a question box, because of what it literally says inside it? Or would you call it a search box, based on our existing context of what bing.com is?
I'm old enough to remember when Amazon's thing was that they were the cheapest. But that old goes so fast. At some point you either raise prices or you lower quality. So, not surprising.
Why use the word "blindly". It looks like it uses a LLM. Possibly one fine tuned on Amazon reviews and product data. It can also answer other questions. Is that bad? Why should Amazon be embarrassed about this?
Because it's a search box, not a question box, and there's no way it's retrained each time somebody posts a review, thus is worse than a database query (arguably, because this is HN).
It explicitly is a question box, and it doesn't need to be "retrained" every time someone posts a review, they're most likely being embedded, summarized, and put into a vector database that the question box can RAG.
This is close to a textbook example of where an LLM is useful. It's interesting seeing just how knee-jerk-y some people's irrational hatred of this has become.
If anything I’m a laggard in getting on the AI hype train, but this seems totally reasonable and actually pretty useful (eg the product I was checking has no less than 11k reviews, so AI giving me a summary answer is also welcome)