Humans Work With AI: A Conversation with FutureProof Music School's John von Seggern

I am currently involved in three projects that need to happen quickly, have a lot of unknowns, and require some serious human collaboration. All three projects are using AI as a processing tool for all the information we need to process. It isn’t replacing anyone, it’s doing some specific functions that would take us weeks to get done otherwise. It’s always flawed, but it’s always useful.That's what I talked about with John von Seggern, who runs Future Proof Music School. We talking about using AI successfully, but using it as a tool, not something that will replace humans.Turning Sticky Notes Into Action PlansI called John for this impromptu chat. At Modus Institute, we run workshops where teams map out their work, their problems, their actions, their collaborations, etc. Everyone puts ideas on sticky notes and puts them on the wall. While these exercises always lead to immediate improvements and have led to faster product development, the actual creation of plans, designs, and schedules takes time.But we tried something new. We grabbed an image and fed it to our AI. The AI read all the sticky notes and wrote up a flawed but useful initial set of requirements. As John says, “It was about 70% right, and that’s great.Rather than using AI to solve our problems for us, we asked it to help us get something done that had a very aggressive deadline. To get there, we had our agent organize the human solutions that were already there. The people in the room did the thinking. AI just helped us initially sort and analyze it faster.Humane Work is reader-supported. To receive new posts and support my work, consider becoming a free or paid subscriber.What AI Does Best (And What It Doesn't)John pointed out something smart: AI is great at "finding patterns in vast amounts of information". Humans can't look at 100 sticky notes and instantly see all the themes. AI can get much closer to this.AI can't replace the human thinking that created those sticky notes in the first place. It can't understand your business like you do. It can't feel frustrated about a broken process or get excited about a new idea.This is because business logic has always been a function of business intelligence and actual intelligence runs on emotion, motivation, and multiple perspectives. While AI claims to have these because its “intelligence” is based on the sum total of what’s on the internet, it’s truly only pattern matching and making sense of a lot of words.Real Examples from a Music SchoolJohn's Futureproof Music School shows how this works in practice. He created an AI music coach called Cadence that teaches basic skills like scales, rhythm, and music theory. But it can’t be creative, it doesn’t have expertise or continuous learning. These are the domain of the human teachers who focus on advanced techniques and helping students with their careers.This makes sense. AI never gets tired of explaining the same concept 15 times. It can (mostly) give consistent feedback on basic exercises. But humans are better at creativity, emotion, and the kind of knowledge that comes from years of experience.They're not competing for the same job. They're working together to do a job neither could do as well alone.An AI Assistant That Actually HelpsJohn as a CEO also built an AI assistant he calls Nova for his business tasks. Nova is integrated with his calendar, chats, and more. Nova looks at all that data helps John run the company. It doesn’t run the company, it keeps track of things and make recommendations. Sometimes they are useful.But the bots only works because John was smart about their limits. He told Cadence to focus only on music and creativity. He made it double-check facts by searching the web. He treats Nova's ideas as starting points, not final answers.This is the pattern we see with successful AI use beyond hype or fear: AI handles the processing, humans handle the judgment. It is a word calculator.The Back-and-Forth ApproachBoth John and I learned that working with AI is like having a conversation. You don't just ask it once and get a perfect answer. You go back and forth, refining your questions and improving the results.As John said, "AI isn't always right, but its suggestions provide a starting point for humans to iterate and refine".This is different from other tools. When you use a hammer, it works the same way every time. AI is more like a thinking partner who might offer useful ideas, obvious points, or occasionally brilliant insights you hadn't considered. It might also be utterly stoned out of its gourd and tell you completely idiotic things.The danger is that often AI will tell you things that look right. They are a plausible story it makes up or takes from sources that were wrong or misinterpreted. AI use does not come with vigilance.How to Use AI WellFrom our conversation, here's what works:Start with human expertise: The best AI applications help people who already know their stuff. The workshop leaders understood team problems. The music teachers understood how to teach. AI just made them faster and more effective.Expect to go back and forth: Don't expect perfect results right away. Plan to refine your questions and improve the AI's responses over time.Set clear limits: Tell AI what it should and shouldn't do. Make it verify facts. Treat its suggestions as starting points, not final decisions.Focus on helping humans, not replacing them: Use AI to handle routine tasks so humans can focus on creative and strategic work.You are also still a stupid human: You can and will use AI to be a confirmation bias engine. You will work your prompts in a way to get what you want, not what you need. Check yourself while you are at it. Sometimes, the AI may be right, it’s not what you want so you think it’s wrong, and correct to your original error.What This Means for Your WorkJohn did not build a music tutor to kill music tutors. He built a search engine, specialized for music. We haven’t replaced our ability to think about how work flows and feature sets are created, we just removed long, confused, and often frustrating processing time between ideation and action. The word calculators are helping move our work forward and get value to people faster, while at the same time allowing us to handle the deeper thinking, human relations, and creative bits.CALL TO ACTION: If this is interesting to you, you can hire me and Toni to help (just message me on substack or linkedin), or take a class at Modus Institute, or simply get a paid subscription to Humane Work here on substack (see above).This is what we do and have done for decades, and what we work with others on. Simple, visual, humane strategies to fix the real problems of work. This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit humanework.substack.com/subscribe