A cryptographic software inserts a detectable signature within the phrases produced by OpenAI’s text-generating synthetic intelligence fashions. It may assist academics cease college students utilizing AIs to do their homework
Technology
13 December 2022
It could be powerful to find out whether or not phrases originated from a text-generating AI or human Manuel Breva Colmeiro/Getty Images
Artificial intelligence agency OpenAI is creating a approach to forestall individuals taking textual content that AI fashions produce and passing it off as their very own work.
The watermark-like safety function may assist academics and teachers spot college students who’re utilizing textual content turbines reminiscent of OpenAI’s GPT to write down essays for them, however cryptography specialists say workarounds will inevitably be discovered.
Scott Aaronson on the University of Texas at Austin, who’s spending a yr working with …