In GUIDE
Quantum computing theorist and computational complexity researcher at UT Austin. Has worked on watermarking methods for detecting AI-generated text during a sabbatical at OpenAI, addressing attribution and authenticity challenges in the era of large language models.
Sources in this wiki
Topics
Notes
Aaronson approached LLM safety from an unusual angle: a quantum computing theorist without prior deep involvement in AI, motivated initially by troll attacks on his blog and concerns about scale of potential LLM misuse. Developed the Gumbel-Softmax watermarking scheme independently, later learning it coincided with foundational work by Kirchenbauer et al.