Humans in Loops, Flows, and Dialogues

Moving to Humans in the Flow

Let’s call that humans in the flow, not humans in the loop. Here’s a flow of the sort in which many of us are already engaged: We’re going on a family vacation to Mexico City. We ask the AI to suggest sites to see. The AI overperforms, as it often does, coming back with not just a list, but also an itinerary that clusters its eight recommended sites by location. But that’s not the flow part. We then throw in that we’re also traveling with three children, ages 10, 7, and 4. The AI recalculates and talks about the suitability of each of the eight sites for the younger children. It rules out four of them and replaces them with ones that would appeal to a wider range of ages.

Great, but then we tell the AI that the youngest one is freaked out by crowds, the middle one wants to know about what ordinary life is like for people in the city, and the 10-year-old is into science fiction. The AI adjusts as we respond to the new possibilities with new information: One would like time in a park, another is afraid of heights, and so on.

We are in the flow, not just responding to the accuracy of the AI’s suggestions, but also, on the basis of those suggestions, refining what would count as a successful visit. That’s a big step beyond being a human checkpoint for the accuracy of the results.

Having a human specialist check the results of an AI diagnostic system is immensely valuable because even if the system achieves a remarkable degree of accuracy, the cost of a mistake can be catastrophic. But the loop we’re actually headed to is much more like a human conversation than a loop. It’s similar in important ways to the conversation you would have with your friend who grew up in Mexico City in that you are educating each other about what would work for your kids. You are diving into the specifics of both the children and the places they might visit. You are discovering possibilities not only of Mexico City, but also of your children. The system is helping you to express the deep tacit knowledge you have about your own kids. And if there’s one thing KM has learned about knowledge over the past 20 years or so, it’s the importance of tacit knowledge.

Humans in Dialogue

In fact, instead of humans in the flow, I think we are entering—possibly are already in—the era of humans in the dialogue with AI, discovering our values, getting more specific about them, and altering their applications based on the specifics of our world and situation.

If the old KM was about building, organizing, sharing, and leveraging knowledge, the new KM might also be about mastering the dialogue: using AI not just to retrieve our answers, but to help us finally articulate the right questions.