This seems like a very good use case for GPT models: Surfacing buried information.

Or to write the press releases published along side SEC filings.

Notbsure why that actually helps, the SEC filongs are public... Justvanother case of the public and media not giving a damn about primary sources anymore.

Edit: From the article:

>> “We find it surprising this behavior is so effective,” Watkins said. “Investors should not assume press releases cover all events happening at a company at a given time.”

Apparently I am not the only one who is puzzled by thale fact the press releases actually seem to work...

I have used it for that very use case while analyzing the terms and conditions of some documents before signing. I still read them all myself as well, because I do not trust the current state of GPT to not miss anything important. But it helped me get a basic understanding and overview before I started my own analysis.

Yes. Indeed. I am working on this for investing. Where I believe the value proposition is the strongest.

I've had this thought before -- that there's lots of good ideas and insights hidden in the deluge of comments on hn, reddit etc -- and GPT might be used to find them. But I haven't figured out a good prompt yet.

This is more of a data question rather than finding a prompt.

You can approach it like that too. I was thinking more in terms of using the existing interfaces.

Sounds like most of the value in that site is locked behind a two thousand dollar a year paywall… wow.

The few articles you can see are interesting though.

All the articles seem paywalled or stubs pointing to articles on other sources. Do you have any readable examples?

The first link is the regular site, which is what I’ve looked at before. For me, much of the value is just in the headline, which flags the issue so I can investigate further if it sounds interesting.