Building a team which contains an AI-Agent
Since we envision the AI agent as an equal member in the team, when building a new team, I interview both people and AI agents…
Q: Wait? Why not use a different AI-tool according to needs?
A: For the same reason you want to work with a human team-member: I want to get to know the agent, and I want it to know me and the team. I want it to understand what’s important to me, and I also want to know the strengths and weaknesses of the AI team member… and vice versa: That it will know my weaknesses, …and the team’s weaknesses.
So just like with a human team-member, I want the AI-agent to be the first address for challenges in the area it is responsible for. But I also want it to be able to say to me: “This is too complex for me. Let’s consult the experts.”
We’re not there yet with AI tools. The following sentence – “Consult another tool on this topic”… is still impossible for AI-tools to express.
Beyond that, I feel a bit like an HR person interviewing 30–50 candidates. I’m trying to build a table of traits, strengths, weaknesses, salary expectations, and so on. The challenge is that from week to week, the candidates improve. And new ones keep coming. Very hard to compare.
So, it is back to me. Facing the following challenge: how do you teach a team member to say “I don’t know”? With a human team member – it’s called Trust. The knowledge that it’s okay to expose their weaknesses. How do you teach that to an AI team member? Maybe by starting every interaction with a statement like this:
“I prefer that when you’re not sure, you tell me ‘I don’t know.’ That will really help me trust your answers.”
So we go back to the beginning: building trust within the team! Trust – that’s lesson No. 1.
“I prefer that when you’re not sure, you tell me ‘I don’t know.’ That will really help me trust your answers.”
So we go back to the beginning: building trust within the team! Trust – that’s lesson No. 1.
And the next question: how can we use an AI tool to help select the candidate. Both human and AI-Agent? That’s hard… There were several attempts to build HR-supporting agents. But the second half of the challenge, seems to introduce a conflict of business interests. But… is it different from consulting with the existing team members when adding a new one? May be. Some team members are sometimes worried when someone strong in their field is brought into the team. Trust? Didn’t we say that?