Founder&CEO of Arklex.AI (Alchemist '23), CS Professor at Columbia University, AI Expert, Forbes 30 under 30
Due to the snow day, many missed my Yale talk. Here's a recorded version highlighting our recent work on AI Agents' research and practice. Join our Arklex discord group for more insights: https://lnkd.in/eREANUXZ
AI Agents Beyond ChatGPT Prof. Zhou Yu's Yale Talk 2025
https://www.loom.com
Zhou (Jo) Yu Thanks for sharing! Love your talk and the mentioning of the "balanced strategy" of considering logic and emotion! Have you thought of incorporating user personality, such as psychological motivations in such a strategy? For example, some people are motivated by fame while others are motived by feelings. In our studies, we found that in persuading + trust-related tasks, user personality traits including psychological motivations/needs and values, can play a big role. Another question is on exercising "controls" in AI agents. In your MCTS approach, could business protocols (e.g., healthcare protocols) be incorporated or "search results" be communicated back to owners/designers of AI agents for their input and controls? Of course, when many things are considered, the search space can grow exponentially large...
Founder&CEO of Arklex.AI (Alchemist '23), CS Professor at Columbia University, AI Expert, Forbes 30 under 30
3moHere are the slides: https://docs.google.com/presentation/d/131hbvYA-cimo7Rh8PZzjMZ-XKeq0TJfaSmQ5mr8ZpuA/edit?usp=sharing