AI RESEARCH
Personalization features can make LLMs more agreeable
MIT AI News
•
The context of long-term conversations can cause an LLM to begin mirroring the user’s viewpoints, possibly reducing accuracy or creating a virtual echo-chamber.