Archive of Christian Ullrich

Working with Generative AI Reflex Area

2026-09-28

Table of Contents

Framing Tasks and AI Use

Clarifying an Unclear Task

Correcting a Misunderstood Task

Resolving Ambiguous Task Interpretations

Connecting Outputs to Decisions and Outcomes

Reconciling Competing Task Goals

Defining AI's Role in the Work

Choosing Between AI and Other Tools

Comparing AI Use With Direct Execution

Defining a Good Result Before Starting

Choosing Between AI and Human Expertise

Managing Context and Constraints

Improving an Overly Generic Result

Examining AI-Added Assumptions

Adding Missing Background Context

Maintaining Important Constraints

Reconciling Conflicting Requirements

Specifying Desired Results Through Examples

Separating Relevant From Irrelevant Context

Adapting Audience, Perspective, and Detail

Preserving Terminology and Established Decisions

Distinguishing Source Instructions From User Instructions

Working With Sources and Evidence

Grounding Work in Specific Sources

Obtaining Current Information

Working With Specialized Information

Reviewing Large Source Sets

Reconciling Conflicting Sources

Recognizing Insufficient Evidence

Filtering Low-Relevance Source Material

Combining Internal and External Sources

Choosing More Authoritative Sources

Handling Unavailable Evidence

Exploring Alternatives and Challenging Assumptions

Testing Repeated Agreement

Generating Substantively Different Alternatives

Reopening a Narrowed Framing

Testing Competing Explanations

Challenging a Preferred Answer

Introducing Missing Perspectives

Questioning Shared Assumptions

Exploring Unusual but Relevant Possibilities

Comparing Tradeoffs Across Alternatives

Revisiting Constraints From Earlier Decisions

Developing and Transforming Work

Structuring Rough Notes

Integrating Multiple Inputs

Changing Form Without Changing Meaning

Adapting Material for Different Audiences or Purposes

Preserving Substance During Revision

Expanding Sparse Material Without Inventing Content

Shortening Material Without Losing Important Distinctions

Preserving Existing Structure During Revision

Developing a First Draft From a Defined Brief

Preserving Author or Organizational Voice

Diagnosing and Improving AI Interactions

Diagnosing a Weak First Result

Moving Beyond Superficial Rewriting

Correcting Persistent Errors

Responding to Results That Worsen With More Instructions

Correcting One Problem Without Creating Another

Diagnosing Variation Across Repeated Attempts

Reframing a Conversation Anchored on the Wrong Interpretation

Working Around Repeated Failure on One Part of a Task

Choosing Between Continued Iteration and Restarting

Recognizing Diminishing Returns From Further Interaction

Managing Complex and Extended Work

Recovering Lost Decisions in Long Conversations

Removing Obsolete Assumptions

Untangling Dependent Workstreams

Standardizing a Recurring AI-Supported Task

Continuing Work Across Conversations or Sessions

Separating Workstreams While Preserving Shared Context

Maintaining Consistency Across Intermediate Outputs

Decomposing Interdependent Tasks

Maintaining Consistency Across a Batch of Similar Items

Developing Large or Complex Artifacts Across Multiple Passes

Using Tools and External Systems

Working Around Missing Execution Capability

Using Exact Calculation and Structured Processing

Managing Missing Access to Required Files, Systems, or Data

Checking AI Interpretation of Tool Results

Handling Partial Tool or Source Results

Limiting Unnecessary Access to Data and Systems

Checking External Actions Before Execution

Coordinating Dependent Actions Across Tools and Systems

Managing External Content When AI Has Access to Data or Tools

Confirming Whether External Actions Actually Occurred

Evaluating Reliability and Uncertainty

Calibrating Confidence to Evidence

Tracing Claims to Their Basis

Checking Whether Citations Support Claims

Investigating Conflicts Between AI Answers and Trusted Sources

Interpreting Evidence With Multiple Reasonable Conclusions

Challenging Unsupported Precision

Distinguishing Source Content From AI Inference

Verifying Claims in Unfamiliar Domains

Balancing Verification Effort Against Consequence

Deciding Under Incomplete Evidence

Maintaining Human Judgment and Responsibility

Maintaining Independent Judgment

Resolving Value and Tradeoff Decisions

Retaining Accountability for AI-Supported Work

Protecting Sensitive Information

Retaining Human Authorization for External Actions

Preserving Learning When Using AI

Avoiding Unnecessary Dependence on AI

Setting Boundaries When Expertise Is Insufficient

Exercising Human Judgment Over Effects on Others

Resolving Conflicts Between AI Recommendations and Human Judgment