|
cat appreciator@kelsbr.bsky.social |
I think these are pretty different situations though. It's not clear to me that OpenAI will be scraping paywalled articles for future GPT models. And GPT-3 was trained on common crawl, which is entirely free and open to the public
1 replies 0 reposts 2 likes
|
Daniel Goldman
@dgoldman.bsky.social
[ View ] |
Besides, much of the actual training data that will go into this model, and even that went into GPT-4o, was synthetic. Besides, we all utilize knowledge we acquire and build on it to create. The issue is simply copying someone else's work and saying "I did this."
1 replies 0 reposts 3 likes