Imagine asking an AI assistant to plan a family gathering. It can suggest a menu, compare prices, draft invitations, and create a schedule. Those tasks may save time. But someone still has to decide what kind of gathering this should be, whose needs matter, and what would make the event meaningful to the people attending.
The same distinction matters in education. AI can take on more and more of the visible work: explaining, drafting, assessing, giving feedback, and adapting activities. The question is not only what AI can do. It is what we want education to help people become and which parts of that work people need to remain involved in.
Education should not try to preserve every task as it has always been done. Some tasks can be supported or automated in ways that make learning more accessible and give students and faculty more time for other work. But the decision to delegate should begin with the purpose of the task. What is the student meant to learn by doing it? What judgment or capability should they develop? What responsibility would be lost if the system handled it entirely?
The answer will differ by course and discipline. If students are learning to write, they need opportunities to make choices about meaning and audience, even if AI helps them revise. If they are learning to solve problems, they need to practice choosing approaches and interpreting results, even if AI offers hints. If they are preparing for work that affects other people, they need to consider consequences and take responsibility for decisions.
This is not a demand that students do everything alone. Human thinking has always involved other people and tools. The important question is whether students remain active in the parts of the work that build their ability to understand, decide, and act.
Research on higher education’s educational purposes points beyond any single skill or job outcome. One review of scholarship on the purposes of higher education found that institutions serve multiple aims, including preparing people for work, supporting personal development, and contributing to society. Those aims can overlap, but they are not interchangeable. Preparing someone to complete tasks efficiently does not by itself prepare them to make sound judgments about which tasks matter or how their decisions affect others. Springer Nature Link
The OECD Learning Compass similarly describes education in terms of knowledge, skills, attitudes, values, and student agency: the capacity to set directions and make choices in a changing world. That broader view matters when AI can generate outputs on demand. Students will need knowledge to evaluate those outputs, judgment to decide how to use them, and agency to choose a direction rather than simply follow the one a system proposes. OECD
UNESCO’s guidance on generative AI in education also calls for a human-centered approach that protects human agency. That principle offers a useful question for colleges: when AI becomes part of a learning activity, does it expand what students can understand and do, or does it quietly take over the decisions through which they learn? UNESCO
What education should refuse to outsource, then, is not a fixed list of tasks. It should refuse to give up its responsibility for helping students develop knowledge, judgment, agency, and the capacity to act responsibly with others. These do not require students to reject AI. They require students to understand when and why they are using it, and to remain connected to the choices and consequences that shape the work.
That is a choice colleges and universities will have to make together. Faculty, students, staff, and institutional leaders will need to decide where AI can contribute, what learning must remain visible, and which relationships and decisions call for human participation. Those choices will shape more than individual assignments. They will help define the kind of higher education we are building.
This series began with questions about teaching, assessment, feedback, classrooms, curiosity, effort, and learning. Taken together, they point to a larger one: What kind of institution do we want to become when AI can do more of the work associated with education?
That is where the next series will begin. The Future of Higher Education will explore the choices institutions face not to predict exactly what colleges will look like, but to consider what we should build intentionally in an AI-rich world.
Continuing the Conversation
Series 1: AI Is Exposing Existing Problems✓ Completed
Series 2: What We Do About It✓ Completed
Series 3: Cultivating Human Thinking in an AI World✓ Completed
Series 4: Learning Alongside AI✓ Completed
Series 5: Preparing Students for an AI World✓ Completed
Series 6: The Purpose of Education in an AI World
Current Post (8 of 8): If AI Can Do Almost Everything, What Should Education Refuse to Outsource?
Next Up: Series 7 — The Future of Higher Education