AI Is Changing Tasks Before It Changes Jobs

Throughout the last four series, I have focused primarily on teaching and learning in an AI-rich world. The first series examined the problems AI is exposing in education. The second considered practical ways faculty can respond. The third explored the forms of human thinking that become more important as AI becomes more capable. The fourth asked how students can learn alongside AI without outsourcing the thinking the learning process is meant to develop. But students will not remain students forever. They will graduate into workplaces where AI is no longer a separate tool they deliberately open when they need assistance. It will be embedded in the systems they use to communicate, analyze information, manage projects, support customers, make decisions, and complete routine work. Some graduates may work alongside AI agents capable of carrying out multistep tasks with limited human involvement. That creates a different question: What should graduates know and be able to do in a world where AI is becoming part of nearly every profession? That is the focus of this new series. Over the next eight posts, I want to explore what it means to prepare students for work that will increasingly involve AI. That includes making AI use visible, taking responsibility for AI-assisted decisions, protecting sensitive information, maintaining professional expertise, recognizing when human oversight is necessary, adapting to workplace policies, and eventually working alongside AI agents. Before we can consider those responsibilities, however, we need a clearer picture of how AI is changing work. Much of the public conversation frames that change around jobs. Which jobs will AI replace? Which careers are safe? Which professions will disappear? What should students major in if they want to remain employable? These are understandable questions, but I am not sure they are the most useful place to begin. Jobs are not single activities. They are collections of tasks, decisions, relationships, responsibilities, and forms of expertise. A nurse does not simply “provide healthcare.” A nurse observes patients, communicates with families, documents care, interprets information, coordinates with other professionals, administers treatments, responds to unexpected changes, and makes judgments under pressure. An accountant does not simply “work with numbers.” An accountant gathers information, applies rules, identifies inconsistencies, assesses risk, communicates findings, and takes responsibility for the accuracy of financial reporting. An instructional designer does not simply “create courses.” An instructional designer analyzes learning needs, works with faculty or subject matter experts, organizes information, designs activities, reviews materials, evaluates accessibility, and determines whether a learning experience supports its intended outcomes. AI will not necessarily affect every part of these jobs equally. It may automate some tasks, accelerate others, introduce new responsibilities, or change how human expertise is applied. In many cases, the occupation may remain while the work within it changes. The International Labor Organization’s 2025 analysis of generative AI exposure estimated that approximately one in four jobs worldwide includes some degree of exposure to generative AI. But the researchers concluded that transformation is more likely than complete replacement because most occupations still contain tasks requiring human involvement. AI may perform parts of the work without being able to assume the entire role or the responsibility that comes with it. (ILO, 2025) We are already seeing evidence of this task-level change. Research involving customer-support workers found that access to a generative AI assistant increased productivity, with the largest benefits appearing among less-experienced workers. The system helped employees respond more quickly and enabled newer workers to perform more like their experienced colleagues. It did not eliminate the role. It changed how some of the role’s tasks were performed and how expertise was distributed. (Brynjolfsson, Li, and Raymond, 2023) A later field experiment involving more than 7,000 knowledge workers across 66 firms found that employees with access to generative AI spent less time on email and completed documents more quickly. Yet the technology did not produce the same change in every activity. AI altered particular work patterns rather than simply making every part of the job faster. (Dillon et al., 2025) This is one reason predictions based only on job titles can be misleading. Two people with the same title may perform very different tasks. Two organizations may introduce the same AI system but distribute responsibilities differently. One employer may use AI to support employees, another to monitor them, and another to automate parts of their work. The technology matters, but so do the choices organizations make about how it is implemented. Preparing students for an AI world therefore cannot mean predicting exactly which tools they will use or which occupations will be most affected. We are unlikely to predict either with much confidence. Instead, students need to learn how to examine work at the level of tasks and responsibilities. Which parts of a job involve producing routine information? Which require interpretation? Which depend on relationships, context, or trust? Which decisions can be supported by AI but still require human review? Which tasks may become faster without becoming less important? What new responsibilities appear when AI enters the process? These questions reveal something that is often lost in conversations about efficiency: when AI performs part of a task, the remaining human work may become more, not less important. If AI drafts a financial summary, someone must determine whether the numbers were interpreted correctly. If it prepares a patient communication, someone must ensure that the language is accurate, appropriate, and safe. If it screens job applicants, someone must examine the criteria, monitor for unfair outcomes, and take responsibility for the process. If an AI agent completes a workflow across several systems, someone must decide what authority it should have, what it should be prohibited from doing, and when it should stop and ask for help. The production of an output may become easier. But evaluating the output, understanding its context, anticipating its consequences, and accepting responsibility for its use may become more significant. This complicates the idea of AI as a simple labor-saving technology. A 2025 OECD survey of more than 5,000 small and medium-sized businesses found that generative AI often reduced workloads and helped some organizations address skill shortages. At the same time, twice as many surveyed businesses said generative AI increased their need for highly skilled workers as said it decreased that need. Data analysis, interpretation, creativity, and innovation were among the skills employers believed had become more important. (OECD, 2025) AI can reduce the effort required to perform certain tasks while increasing the importance of the expertise needed to supervise, interpret, and apply the results. That has important implications for how we talk with students about career readiness. Telling students simply to “learn AI” is not enough. The tools they learn today will change. Features that currently require specialized prompting may soon be built into the ordinary software used across a profession. Students may rarely be asked whether they know how to use a particular chatbot. They may instead be expected to work effectively in an environment where AI is already present. In that environment, technical familiarity will matter. But so will knowing the work well enough to understand what the AI is doing. A student who can generate an answer but cannot recognize when it is inappropriate is not prepared. A student who can automate a process but cannot identify the risks created by that automation is not prepared. A student who can produce polished professional work but cannot explain how it was created is not prepared. A student who can use AI effectively only when the tool is available has not necessarily developed professional capability. This does not mean students should avoid AI so they can prove they can do everything independently. Professional work has always involved tools, systems, colleagues, and forms of distributed expertise. Refusing to use appropriate tools is not the same as demonstrating competence. But neither is using a tool successfully. Professional competence increasingly includes knowing what to delegate, what to retain, what to inspect, and what responsibility remains with the human even after part of the work has been handed to a machine. This may be one of the most important ways AI is reshaping work. It is changing not only how tasks are completed, but also where human attention needs to be directed. When producing a first draft becomes easier, reviewing may become more important. When information becomes easier to summarize, identifying what was omitted may become more important. When recommendations arrive faster, questioning their assumptions may become more important. When processes become automated, understanding who remains accountable may become more important. When AI systems begin taking actions rather than merely producing content, setting boundaries and maintaining oversight may become more important. Preparing students for this world requires more than adding an AI tool to an existing assignment. It requires helping them see professions as systems of tasks, judgments, relationships, and responsibilities. Students need opportunities to examine where AI could contribute to work in their fields. But they also need to consider what changes when it does. What becomes easier? What becomes harder? What new risks are created? What knowledge is still necessary? What human capability becomes more valuable? Who remains responsible when something goes wrong? These are not questions with one answer. The answers will vary across professions, organizations, and situations. That is precisely why students need practice asking them. The future of work will probably not arrive as a single moment when entire occupations suddenly disappear. It is more likely to emerge through thousands of smaller decisions about which tasks are automated, which are augmented, which are redesigned, and which remain firmly under human control. Our graduates will participate in those decisions. Some will choose the tools. Some will implement the systems. Some will supervise their use. Some will be affected by decisions made by others. Nearly all of them will need to determine what responsible professional practice looks like when AI is part of the work. Preparing students for an AI world, then, is not simply about helping them become more efficient. It is about helping them understand how work is changing and what responsibilities they will carry within that change. And one of the first responsibilities will be making clear when and how AI contributed to their work. That is where the next post will turn. 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
Current Post (1 of 8): AI Is Changing Tasks Before It Changes Jobs
Next Up: Making AI Use Visible in Professional Work‍