AI Is Exposing the Managers Who Stopped Learning

AI Changes the Knowledge Required to Manage

AI has changed the knowledge required to manage software engineering, and too many managers are still trying to lead with knowledge acquired before that change.

For years, managers could get away with this. As engineers moved into management, they moved away from the technology itself. Their calendars filled with 1:1s, planning sessions, status updates, hiring, processes, roadmaps, and alignment meetings. Past experience carried them surprisingly far, so there was little immediate penalty for no longer systematically learning how the work itself was changing.

AI breaks that bargain. There is no established blueprint that managers can simply adopt, because the technology and the practices around it are still evolving. Managers therefore need to learn again. Not because they should become the best AI users on their teams, but because management is decision-making and decisions are only as good as the knowledge behind them.

Managers decide what work gets done, how teams work, which constraints apply, where money is invested, what quality means, who is performing well, and what should change. If their understanding of AI is weak, those decisions are increasingly being made with missing knowledge.

This matters because knowledge is the constraint of knowledge work. AI has changed some of the knowledge required to manage that system. A manager who has never seriously used AI cannot easily judge where it creates real leverage, where it creates risk, what good AI-assisted engineering looks like, or what organizational conditions allow it to work.

The problem gets worse as you move up the hierarchy. An individual contributor who does not understand AI primarily limits their own leverage. An engineering manager can limit an entire team. A director can limit several teams. A VP of Engineering can make decisions that constrain hundreds of people. Organizational authority multiplies the cost of missing knowledge.

That is why companies may be making a mistake when they focus their AI-adoption pressure primarily on individual contributors. The people with the greatest organizational leverage should face at least as much pressure to learn and probably more.

The higher you climb in engineering leadership, the more expensive your AI ignorance becomes.

Your AI Ignorance Becomes Everyone Else's Constraint

When managers do not understand AI, their ignorance propagates through every decision they make and becomes a constraint on everyone they lead.

The first casualty is judgment. A manager who lacks firsthand AI experience struggles to distinguish real leverage from AI theater, disciplined AI-assisted engineering from sophisticated-looking output, or a genuine automation opportunity from wishful thinking. They may push AI where human judgment is essential while preserving manual processes where AI could remove hours of unnecessary work.

Developers notice this gap quickly. Consider the manager who asks, “Can't we automate this with Claude?” without understanding the work being discussed. The problem is not the question itself; sometimes the answer really is yes. The problem is asking it without enough knowledge to understand the answer. The manager who should be creating leverage has instead become another knowledge gap the team must compensate for.

That gap eventually becomes a trust problem. Strong engineers will initially work around a pre-AI system. Then they will become frustrated with it. Eventually, some will leave for organizations where management enables their leverage instead of suppressing it.

The damage compounds: AI-ignorant leadership leads to bad decisions, constrained developer leverage, loss of trust and talent, weaker delivery, and eventually competitive decline.

One developer who fails to adapt can underperform. One senior leader who fails to adapt can preserve obsolete processes, make poor investments, reward the wrong behaviors, block useful experimentation, and constrain hundreds of people who have adapted.

Nothing destroys confidence in leadership faster than discovering that the person telling you how to work no longer understands how the work can be done.

Use AI. Learn From Your Team. Then Redesign the System.

Managers do not need to become the best AI users on their teams, but they do need enough firsthand understanding of AI to make good decisions about the people and systems they manage.

The answer is not another mandatory course followed by a certificate declaring managers “AI ready.” AI competence comes from building intuition about what the technology can and cannot do in real work. That requires experience: trying to solve actual problems with AI, seeing where it surprises you, discovering where it confidently fails, learning how much context changes the result, and experiencing where human judgment remains indispensable. Personal AI use is the learning mechanism; management judgment is the capability you are trying to build.

There are three steps, and they should happen as a progression:

Use AI yourself: Protect time every week to put AI to real use. Build something, research a decision, automate a repetitive task, analyze information, or use an agent to complete work you previously did manually.

• Benefit: You develop intuition that no presentation or vendor demo can give you.
• Risk: Personal productivity can create false confidence if you assume your use cases represent engineering work.

Learn from the people doing the work: Sit with your strongest AI-using ICs. Ask what AI makes easier, what remains difficult, where it fails, what context it needs, and what organizational constraints prevent more leverage.

• Benefit: You replace assumptions with knowledge of what is actually happening inside your teams.
• Risk: You must be comfortable discovering that people who report to you understand an important part of the transformation better than you do.

Redesign the system: Change workflows, specifications, architecture practices, quality controls, tooling, roles, governance, incentives, and expectations where AI has changed what good engineering looks like.

• Benefit: You turn individual productivity gains into an organizational capability.
• Risk: The uncomfortable discovery may be that your management system is the main thing that needs to change.

For an IC, adoption may mean becoming substantially better at performing work with AI. For a manager, that is only the beginning. Companies should hold managers to a higher, not lower, standard as organizational leverage increases.

Use AI yourself. Learn from the people doing the work. Then redesign the organization around what you discover.

Personal AI use is the classroom. Better management judgment is the objective.

AI Literacy Is Becoming a Test of Management Relevance

AI literacy is becoming a test of management relevance: managers who learn can multiply the leverage of their organizations; managers who refuse to learn will increasingly become the constraint.

Managers who understand AI can recognize where it genuinely changes the economics of work, distinguish useful experimentation from AI theater, and have credible conversations with the people doing the work. More importantly, they can turn what individual engineers discover into organizational capability by changing processes, removing constraints, spreading effective practices, and redesigning how teams operate.

Managers should not fear discovering that their ICs know more about AI than they do. They should exploit it. The manager's job is not to possess more knowledge than every person on the team; it is to ensure that the knowledge distributed across the team results in better decisions and better organizational performance.

Do nothing, and pre-AI workflows survive because nobody with authority understands why they should change. AI-capable engineers become trapped inside processes designed around old assumptions. Employees notice the contradiction when management asks them to use AI more while preserving the constraints that prevent meaningful gains.

The differentiator becomes whether a management system learns faster than its competitors—whether leaders can convert what AI makes possible into new ways of building software. AI competence should increase with organizational leverage.

If a manager does not understand how the work is changing, cannot evaluate how effectively people are using the new technology, cannot identify where it creates leverage, and cannot redesign the system around those capabilities, then what exactly is the organization receiving in return for that manager's authority?

The hopeful part is that this outcome is not inevitable. Managers can learn, experiment, ask their ICs to teach them, rebuild their intuition, and use their authority to remove constraints. The same management leverage that makes ignorance dangerous makes an AI-literate manager extraordinarily valuable.

AI will not make engineering managers irrelevant. Refusing to learn what AI changes might.

Next Step

Hold managers to a higher AI standard than ICs: give them time to learn by doing, expect them to learn from their teams, and hold them accountable for turning that knowledge into a better engineering system.