simonw 2 days ago

I'm always interested in seeing which definition of "AI agents" a group is using (which they inevitably seem to assume is obvious and shared by everyone else.)

In this case it looks like it's "AI decision-making system that learns through reinforcement learning to handle complex tasks" - with training an algorithm to control traffic lights at many intersections in a city as their illustrative example.

  • falseprofit an hour ago

    What alternative definitions(s) have you found to be most common?

j45 2 days ago

In case anyone may be researching lately, I'd appreciate learning about potentially any useful content, tools, or suggestions to build or train more AI agents.

CrewAI is already on the radar :)

  • d4rkp4ttern a day ago

    Have a look at Langroid [1], a multi-agent LLM framework from CMU/UW-Madison researchers (I am lead dev). It does not use any other LLM library (CrewAI uses LangChain). We started building Langroid in Apr 2023 as an agent framework from the ground up, with a particular focus on an agent orchestration mechanism that seamlessly handles inter-agent hand-off as well as Tool/Function-handling.

    [1] https://github.com/langroid/langroid