Reddit CEO Steve Huffman said Reddit would start charging for using its application programming interface (API) because AI companies, such as Google and OpenAI, had been using the site's data to train their large language models...
This is so strange to me. Doesn't he know that they already scraped the site? That ship has sailed. They don't need to continually scrape reddit. You don't back an LLM with a live website, you use an offline repo of the data.
I still think 17+yrs of internet bickering isn't too very different from 18 or 19. If I got 17 for free, do I feel the need to pay millions for each year in addition?
I feel like the way humans communicate is probably able to be gleaned by an LLM bot on even a single year of data.
I'm probably missing the big picture though. I seem to recall reading that with LLM's, the "L-er" the better.