CT in the AI Era
How large language models and AI agents reshape computational thinking — what changes, what gets commoditized, and what becomes more valuable than ever.
For fifty years the bottleneck in computing was expressing solutions: you had an idea, and the hard part was translating it into working code. Large language models have inverted that bottleneck. Generation is now cheap and nearly instant; the scarce skills are deciding what to build, specifying it precisely, verifying that what came back is actually correct, and knowing which parts of the work should never have been delegated in the first place.
That inversion does not make computational thinking obsolete. It redistributes its weight. Decomposition resurfaces as problem formulation and specification writing. Evaluation and debugging — historically the neglected pillar — becomes the dominant daily activity. Abstraction now includes holding an accurate mental model of a stochastic text generator. This section maps the classical pillars onto AI-era practice:
| Page | What you will learn |
|---|---|
| Problem Formulation | Why formulating problems and writing specifications is the new decomposition, and how to do it well |
| Verification Thinking | Why evaluation skills now dominate generation skills, and practical strategies for verifying AI output |
| Human–AI Division of Labor | A decision framework for what to delegate to AI agents and what to keep human |
| Model Thinking | A practitioner’s mental model of how LLMs work — tokens, probability, context windows, hallucination — with no math |
| Enduring Skills | Which CT skills matter more in the AI era, and which have been commoditized |
The through-line
Every page in this section rests on one claim: the value of a skill tracks how hard it is to verify, not how hard it is to produce. Anything a model can generate and you can cheaply check has been commoditized. Anything that requires judgment to specify or evaluate has appreciated. Reading this section will not teach you any particular AI tool — tools churn quarterly. It teaches the thinking layer above the tools, which is the part that transfers.