In a demo Q&A for Ars Technica hosted by OpenAI, the company’s representatives would not specify exactly if the search info came from Microsoft’s Bing search engine or its own crawled search index, saying it pulls from a “blend” of sources, including partner media organizations.
It appears to be a new Bing reseller.
One of the first searches I tried was an attempt to find torrents for a show. ChatGPT blocked the attempt the best it could, directing me to “free and legal options” like Internet Archive.
It was able to easily find the Steamed Hams Umineko game.
So far, it seems most effective for finding where to watch a movie or show legally.
When looking for manga to read, it straight up links me to fan scanlations.
In the Ars technica piece (thank you @gnome) “hopes” is doing a lot of heavy-$$-lifting:
“OpenAI hopes the new capability will streamline web searching by eliminating the need for multiple searches and link exploration that traditional search engines sometimes require.”
The interview is good insight if you don’t know how much OpenAI has changed. Thanks and for the time stamp @itsMe; I didn’t listen all the way through so I’ll add these posers in case not covered. Who recently invested in OpenAI and who has joined the board since the “coup”? The answers will tell you who OpenAI is in bed with.
And thank you @bbbhltz for the links to Emily’s papers; I’d not read the 2024 one. Her work is invariably excellent, and based on a scan and the abstract it’s a must read.