Another "Ah-ha" Moment in Generative AI

I recently had?another "ah-ha" moment in Gen AI (the first one was after using ChatGPT), when watching a bunch of AI agents?collaborating autonomously, with no human interaction, to analyze, strategize and solve a complex problem.

Using today's technology, you can build AI agents with the following capabilities:

1. Conversation: Receive?user requests, get clarification, request additional information, and provide responses
2. Strategy: Analyze the problem and formulate a strategy
3. Workflow: Follow a multi-step process, including conditionals and retracement
4. Tool use: Make use of user-provided code functions; independently write, review and execute new code
5. Data access: Retrieve data from APIs and databases; read and write files
6. Collaboration: Autonomously communicate with other AI agents without human interaction
7. Inner voice: Internal monologue(chain of thought) to think about the problem, formulate an approach, track progress and make course corrections until reaching the goal

While each of these capabilities is remarkable in itself - none of this was easy to implement even a couple of years ago, without large language models - taken together I believe this approach represents a major step forward in AI capabilities.?

Just as an average team of humans, composed of managers, generalists and specialists, can typically outperform any single human, however brilliant - in the same way, a team of AI agents with different focus areas, built with lower-level models and working together collaboratively, are likely to outperform individual requests to much higher-level models; alternatively, this approach will enable the best models to do more, implementing challenging tasks and solving ever more complex problems.