Once a week, I take an electrician's class. It's a hedge against AI obliterating the white-collar economy. At 39, I still have at least 25 years before retirement, and I'd like to acquire an AI-proof skill that could provide a decent living should the economy become unstable.
My instructor is dreadfully boring. He's a master electrician, undoubtedly skilled at solving any electrical problem. But his grasp of physics is weak, and his instructional methods range from reading the textbook to reading the slides. I get far more enjoyment from exchanging with ChatGPT about magnetic fields and parallel circuits — its explanations as complex or elementary as I wish — than listening to the instructor read off bullet points. If I don't understand something, I do not raise my hand. Instead, I make a note to later ask ChatGPT. Why bother admitting to peers that I don't understand something when I can keep it to myself and get a better explanation?
This experience with AI, among many others, has convinced me that AI can provide excellent instruction and accelerate development. However, it can also erode authority, leading the user to rely on AI more than human experts. As AI continues to advance, people will increasingly turn to it as their primary source of authority or, at least, a check on it.
Despite its risks, lean leaders and managers should be open to AI to scale their teams' development. Lean coaches are overloaded. At LEI, we see large corporations tasking tiny continuous improvement functions with "rolling out lean" to thousands of people. These CI leaders are set up for failure. They simply do not have the time to adequately develop problem-solving capability in anyone when asked to do so for everyone. The consequence is watered-down training that is effective at checking corporate boxes but ineffective at developing people.
AI has the potential to provide exceptional problem-solving coaching to limitless people. How many can a CI team of five directly impact? A few dozen? Certainly fewer than a hundred employees. A well-trained AI has no limit, and it can run 24/7.
Moreover, AI can offer highly critical feedback without being intimidating. Sharing an A3 can be nerve-racking. For one, it's your thinking, highly visualized. There is no place to hide. Secondly, thinking is visualized so your manager can critique it. Early drafts can earn harsh words. Rapidly revising those early drafts with an AI's coaching is simply more enjoyable than getting tough words from a manager.
Managers also benefit. They'll receive more competent A3s, allowing them to spend less time correcting fundamentals and more time aiding complex problem-solving.
AI can also be trained to meet users where they are. Too often, lean coaches take a one-size-fits-all approach to coaching: Socratic. No matter the problem-solver's aptitude or experience, they rely on open-ended questions. This works if you trust the technical skills of your coach. At LEI, I've had the privilege of working under experienced coaches who would reply to my questions in frustrating ways. For instance, one often responded to my suggestions or questions with a frowning, "Really?!" and then walked away. He had previously demonstrated such enormous technical know-how that I not only trusted this method but took it as a show of respect — he thinks I can figure this one out!
However, I've also encountered open-ended questions from coaches who clearly lacked expertise to ask a more illuminating question. I never again sought their help. Coaches earn the right to ask such questions through the acquisition of expertise and experience. Without expertise, how can they know whether the learner is heading in the right direction? How can they help at all?
People are much more willing to ask for help when the help they get is actually helpful. A well-trained AI coach can do just that by providing situational coaching that matches the user's needs. Young employees adept at AI will quickly grasp their manager's technical knowledge, consequently informing the level of trust they place in them. If low, employees may rely heavily on AI as their primary authority to direct their work and problem solving. And that's a problem.
AI has already become an extension of young people because of its pervasive use in college. To grasp the scale of its use on campuses, head over to TikTok. Students have posted innumerable videos explaining how to produce everything from voice assistants that tutor organic chemistry to undetectable AI-generated essays. Its use is so common that students not using AI share strategies to prove innocence in case of a teacher's false accusation. And the pace of development is so rapid that students share stories about "back in the day", which was no more than 14 months ago, when sophisticated tools to conceal their use of AI were unnecessary.
As students graduate, companies need to adapt to their learning styles. AI is already foundational to their development. If their employers do not provide AI tools that assist their development, they are capable of designing it themselves, without any connection to human teaching and coaching. Companies need to decide whether they want to design and complement the tools or hand that responsibility entirely over to their employees.
Try an experiment! Design a problem-solving coaching agent and trial it with your team. If you're interested in joining the beta for LEI's AI problem-solving coach, sign up here.
Matthew Savas is a lean practitioner and writer at Lean Enterprise Institute, exploring the intersection of AI and lean thinking.
Get articles like this delivered to your inbox. Join technology leaders learning to apply lean thinking.