Why is everyone outraged when Google/Microsoft/Yahoo and others have scraped the whole internet for two decades and are also massively profiting from that data?
There’s a significant difference in the purpose of the scraping.
Google et al. run crawlers primarily to populate their search engines. This is a net positive for those whose sites get scraped, because when they appear in a search engine they get more traffic, more page views, more ad revenue. People view content directly from those who created it, meaning those creators (regardless of whoever they are) get full credit. Yes, Google makes money too, but site owners are not left in the cold.
ChatGPT and other LLM’s works by combing its huge database of known content its “learned” to cook up an answer through fast math magic. Content it scrapes to populate this database can be regurgitated at any time, only now its been completely processed and obfuscated to an insane degree. Any attribution of content is completely stripped in the final product, even if it ends up being a word-for-word reproduction. Everything OpenAI charges for its LLM goes directly to OpenAI, and those who have created content to train it will never even know it was used without their consent.
Essentially, LLM’s operate like a huge middle school plagiarism machine shitting all over any concept of copyright, only now they’re making billions off said plagiarism with no plans to stop. It’s a huge ethical conundrum and one I heavily disagree with.
Why is everyone outraged when Google/Microsoft/Yahoo and others have scraped the whole internet for two decades and are also massively profiting from that data?
There’s a significant difference in the purpose of the scraping.
Google et al. run crawlers primarily to populate their search engines. This is a net positive for those whose sites get scraped, because when they appear in a search engine they get more traffic, more page views, more ad revenue. People view content directly from those who created it, meaning those creators (regardless of whoever they are) get full credit. Yes, Google makes money too, but site owners are not left in the cold.
ChatGPT and other LLM’s works by combing its huge database of known content its “learned” to cook up an answer through fast math magic. Content it scrapes to populate this database can be regurgitated at any time, only now its been completely processed and obfuscated to an insane degree. Any attribution of content is completely stripped in the final product, even if it ends up being a word-for-word reproduction. Everything OpenAI charges for its LLM goes directly to OpenAI, and those who have created content to train it will never even know it was used without their consent.
Essentially, LLM’s operate like a huge middle school plagiarism machine shitting all over any concept of copyright, only now they’re making billions off said plagiarism with no plans to stop. It’s a huge ethical conundrum and one I heavily disagree with.