3 ms·
Reddit battles OpenAI and sees its pre-IPO valuation fall
- flangola7 3y agoChatGPT is already trained though, the previous 17 years are already in the datasets. Horse left barn a long time ago. Ending the API doesn't even do anything. Chrome and Edge can simply index every page users visit. OSS models can simply use scraping bots. You can't have a public webpage and also prevent it from being copied. That's now how information science works.
- nullc 3y agoYou can, however, make the site increasingly useless to ordinary users that don't have a team of experts to work around your changes. For example, Reddit has been quite difficult to search for a couple years now. Their logged out views salt pages with large amounts of hidden text from random unrelated discussions, so when you google for a specific string in some old comment there is good odds that the page you click on won't even have visible (or logged in, have at all) the text you were searching for. I'd be mystified that that kind of bad conduct hasn't gotten them outright banned from search indexes, but the fact that reddit is one of the few resources that keep search engines useful (even in its sorry state; the above problem mostly hurts comments not posts) might have something to do with it. In any case, it's a common pattern: the intended target of your mitigation can absorb it as a cost of doing business, the collaterally damaged users cannot.