Now that AI can write reasonably good-sounding copy, reviews are increasingly unreliable.
Comment on LegalEagle Suing PayPal's Honey
dependencyinjection@discuss.tchncs.de 5 days agoI don’t trust reviews at all at this point, from any service like those mentioned.
I will say that it’s diabolical that trust pilot closed the reviews. Meaning people can’t express there disappointment with the app, and that people might still trust it.
rumba@lemmy.zip 5 days ago
RunawayFixer@lemmy.world 5 days ago
Aggregate scores on all sites have become untrustworthy, they’re just poor first indicators now, but reading user reviews is still very much worth it imo. It just takes way longer to figure out whether a product is good/bad than it did 10 years ago. Once ai llm catch up with writing credible texts, then that method will be toast as well and then we’ll be really screwed when choosing a product.
And I kinda understand why they’re blocking new reviews. Trustpilot doesn’t have a way to verify if the reviewers are actual product users, so their system is very vulnerable to review bombing. It’s a catch 22 for them: damned if they suppress review bombs and damned if they don’t.
Trustpilot’s method could be better (Fe: they could allow reviewbombs to happen and show 2 scores, with and without), but what Google is doing is probably the worst possible way to go about it: On the chrome webstore page there is no indication whatsoever that anything is amiss. Atleast Trustpilot tells visitors to go check the news.
I actually can’t believe that I’ve been defending Trustpilot, they’ve always had a repuation of selectively removing reviews, but well, Google is now worse than them.
sugar_in_your_tea@sh.itjust.works 5 days ago
It just takes way longer to figure out whether a product is good/bad than it did 10 years ago
Exactly.
On Amazon, I need to go at least a couple pages in to get past the “curated” comments or whatever to see legitimate reviews. I try to sample a dozen or so starting from at least a couple pages in to see what the trends are (and no, I don’t trust the AI crap Amazon shows), and I’ll read through some 2-star reviews as well (1-star reviews seem to mostly be complaints about shipping or defective products, which may not be relevant).
ai llm catch up
We’ll just have to go back to how we used to do things: word of mouth. In the modern age, social media can absolutely help, provided you trust the author. I have some YouTube channels I trust, some websites that haven’t yet been overtaken by AI nonsense, etc. And the last option we have is returning bad products, and most companies seem to have automated returns to the point where you don’t really need a good reason to return something, it’s generally cheaper for them to accept the return than to piss off their customers.
rumba@lemmy.zip 5 days ago
Once ai llm catch up with writing credible texts,
We’re there. Current-gen stuff is good enough you’d have no idea. Kind of a catch-22, once it’s that good, there’s no way to tell it’s that :)
faultyproboscus@sh.itjust.works 5 days ago
Trustpilot tries to weed out fake reviews. A huge influx of reviews all at once looks like fake reviews. And, to be fair, I imagine a chunk of those reviews are “fake” in that the reviewers never used the app. It’s easier for Trustpilot to cut off new reviews for the time being than to deal with evaluating all these new reviews.
dependencyinjection@discuss.tchncs.de 5 days ago
Wouldn’t you expect a large influx of negative reviews when news breaks of this story?
As I said I don’t trust Trust Pilot, but this really doesn’t help their cause.
sugar_in_your_tea@sh.itjust.works 5 days ago
Sure, and I’d expect a lot of those to be from users who don’t use Honey, but are outraged by the news.
If they were going to leave a legitimate negative review, they would have done so before the news broke.
dependencyinjection@discuss.tchncs.de 5 days ago
I don’t think they would know about the issue until the news broke. The average user would assume it is doing what it said, and the content creators were non the wiser either.