The biggest problem facing AI these days? It's running out of things to read.
AI bots like ChatGPT are "starting to hit a brick wall" when it comes to finding new text to help the tech continue to improve, according to Stuart Russell, a professor at the University of California at Berkeley who is an expert in the field.
Russell's revelation, which came while he spoke at a recent UN summit, is foreboding news for the cutting-edge tech. Text is the lifeblood of AI chatbots, helping them better understand what to say and when to say it.
As if that's not bad enough, the pool of text that chatbots can pull from for training might be shrinking as well.
A new lawsuit alleges Google has been "secretly stealing everything ever created and shared on the internet by hundreds of millions of Americans" to train its AI products.
Clarkson Law Firm, which filed the lawsuit, also lodged a similar complaint against OpenAI in late June. The complaint described the way the startup collected data to train its chatbot, ChatGPT, as "the negligent and otherwise illegal theft of personal data of millions of Americans who do not even use AI tools."
It's not just personal data that has people up in arms. Authors are arguing that using their published books to train chatbots is a violation of copyright law.
That argument could be difficult, though, because chatbots aren't simply redistributing copyrighted works, writes John Eden, a lawyer who is an expert in copyright law.
"These platforms are rapidly learning from large amounts of publicly accessible data and then generating new content," Eden wrote.
So how does this all end?
Eden has some theories, including AI companies adopting and publishing data-use policies. But in the short term, one thing seems certain: more litigation.
Media companies have already indicated they're ready to go to war to protect their content from AI chatbots if they aren't compensated for it.
In the meantime, tech companies probably won't look to slow down.
In response to the lawsuit, Google's general counsel, Halimah DeLaine Prado, previously told Insider that "American law supports using public information to create new beneficial uses" and that "we look forward to refuting these baseless claims."
No comments:
Post a Comment