Quote:
Originally Posted by ratinox
Which is great until they tell you outright lies. Over the weekend, ChatGPT's summaries insisted that the US invasion of Venezuela didn't happen. You can't trust the clickbait titles because they're designed to trick you, and you can't trust the thing that's supposed to cut through the clickbait titles because they're designed to show you what the operators want you to see rather than the truth. Which is to say, they're also designed to trick you.
|
Possibly. But at least from what I've seen, it did a decent enough job to let me deduce what the article was about, and make a decision about whether to read it.
For example, here's a clickbait article that showed up in my news feed:
The Surprising Reason This Obscure Star Wars Game’s Price Has Quadrupled
Meanwhile, the Google News AI summarized it as (paraphrasing):
"Star Wars: Racer Revenge’s has skyrocketed in value because it contains a potential exploit for hacking the PS5."
That's not a classical newspaper headline, but it more closely follows
the spirit of old headlines, which is to convey information, and let
me decide whether I want to read more about it (I would), rather than the clickbait method of "Titillate the reader and withhold all useful information until they open the article."