Title:  Artificial Superintelligence (ASI) : AI agents and Super-alignment

Speaker:  Jingfeng Yang  ( Scientist at Amazon)

Time:   10:00AM, Friday 19th July

Location: 106, WICT

Abstract:  Although artificial general intelligences seems to be more achievable with the power of LLMs, there are still tremendous challenges towards Artificial Superintelligence. In this talk, we will first explain why AI agents and superalignment are key to artificial super-intelligence. Then we will discuss core challenges to develop autonomous agents and possible solutions to them. Finally, we will discuss some approaches to achieve superalignment.

Bio:  Jingfeng Yang is a scientist at Amazon. He is a founding member to develop Amazon Rufus, a general AI Chatbot at Amazon App aligned as a shopping assistant, where he contributed to LLM pretraining, alignment, evaluation, and agents. Prior to that, he graduated under the supervision of Prof. Diyi Yang, and did research at Google and Microsoft. He regularly published at top AI/ML/NLP/Data Mining conferences. He also wrote popular blog posts regarding LLMs, AI safety and alignment.