Your agency is deploying AI. Tasks are getting faster. And somewhere downstream, something is quietly breaking.

That is not an AI problem. It is a systems problem. AI is exceptional at trees. It drafts, codes, summarizes, and responds at the task level. What it cannot do is see the forest: the interconnections, delays, and feedback loops that determine whether your agency actually delivers better outcomes. When you optimize one node in a system, the constraint does not disappear. It moves. If you cannot see where it moves, you end up doing the wrong thing faster.

This session makes the case that systems thinking is now the defining skill for IT knowledge workers. As AI absorbs task-level work, your value shifts up a level to system-level judgment. You will learn to spot the local optimization trap, read the feedback loops and delays that hide in government processes, and use the iceberg model to move from firefighting events to fixing structure. You will leave with three practical tools you can apply Monday morning, no diagramming software required.
AI will keep getting better at the trees. The people who can see the forest will decide what all that speed is for.

Learning Objectives:
After this session, participants will be able to:
1. Identify the local optimization trap in AI deployments and other process improvements, and predict where a constraint will move next when a single node in a workflow is accelerated.
2. Map the basic feedback loops and delays in a government process using simple causal loop sketching, without specialized software or formal modeling training.
3. Apply the iceberg model of events, patterns, and structure to distinguish firefighting from structural fixes, and reframe recurring problems as system design questions rather than performance issues.
4. Use three practical systems thinking tools, including the five whys upstream habit, constraint tracking, and leverage point questions, to evaluate where AI and automation will genuinely improve outcomes versus simply shift the bottleneck.