A new paper from OpenAI has shown why a little bit of bad training can make AI models go rogue—but also demonstrates that this problem is generally pretty easy to fix. Back in February, a group of researchers discovered that fine-tuning an AI model (in their case…
Read More
