AI Assistance Raises Questions About Motivation and Skill Retention

When AI support disappears

The debate over AI’s effect on motivation has a sharper focus: people who used AI while solving math problems or reading comprehension tasks were more likely to struggle or quit when that support disappeared. The findings, presented earlier this month at the 2026 Conference on Language Modeling in San Francisco, suggest that assistance can change how people approach difficult work after the tool is gone. That’s a narrower claim than saying AI makes people lazy: it concerns what happens when a particular kind of help is withdrawn after people have used it during a task.

That distinction matters because performance with support and persistence without it are different measures. A person may finish an exercise faster with AI and still have trouble continuing alone; another may use help to understand a concept and then work independently. The reported finding points to a possible dependence effect, but the article doesn’t provide details about the study’s design, the size or composition of its participant group, or how researchers measured struggling and quitting. Those limits make it difficult to tell how broadly the result applies. They also mean the finding should be read as a reason to examine how support is used, rather than proof that AI inevitably weakens motivation.

The study’s results prompted Live Science to ask readers whether AI leaves them less motivated. More than 140 people voted. The largest reported group, 30%, said the question didn’t apply because they don’t use AI. That result is useful as a snapshot of this poll’s respondents, but it doesn’t establish how common AI use is among readers generally, or whether the people who voted reflect a wider population. A poll about personal impressions can show what participants chose to report; it can’t, by itself, explain why they feel that way.

Concerns about attention and effort

Some readers saw the poll as part of a broader concern about dependence on automated tools. One commenter argued that AI can make people lazier by providing answers without requiring them to search or work through a question. The same commenter also raised claims about AI and personal safety, though the poll itself focused on motivation. That separation is important: broad concerns voiced in comments aren’t evidence about what the study tested, and the safety claims don’t tell us whether AI changes persistence on academic tasks.

The underlying worry is familiar across tools that supply answers or reduce steps in a task: when a person can skip searching, comparing, or working through an explanation, they may get a result without practicing the process that would help them do it unaided. But the tool’s presence alone doesn’t show which parts of that process a user skips. AI might provide a shortcut to an answer, or it might help someone get past a confusing point and continue. The effects depend partly on the task and on whether the user engages with the reasoning behind the response.

Another reader compared AI summaries with navigation devices, saying that relying on them can weaken attention and retention. In that reader’s view, AI can speed up research, but frequent use of summaries may also reduce the effort people put into evaluating information for themselves. The comparison gets at a practical trade-off: a summary can make a long text easier to scan, while also leaving the reader with less direct contact with its detail and argument. Here’s the thing: saving time isn’t the same as learning more. Whether a summary supports understanding depends on what the reader does with it afterward.

Tool use or reduced motivation?

Other commenters challenged the idea that AI itself drives disengagement. One asked whether people would respond the same way if answers to most problems were printed on a separate sheet. Another compared AI to a shovel: a tool that changes how a task is done, without necessarily changing someone’s willingness to do it. The analogy is useful, though it doesn’t settle the question. A shovel helps perform physical work; an AI answer can sometimes supply the result or explanation that the user would otherwise have had to produce.

That difference makes the form of assistance worth examining. A tool that offers hints, asks a learner to explain a step, or helps locate a relevant passage may leave the core work with the user. A tool that supplies a finished answer can remove more of that work. These are general possibilities, not details established about the study described here. Still, they show why “using AI” is too broad a category to explain a change in motivation on its own.

That distinction matters. The study points to a possible cost when support is removed, while the reader responses raise a harder question: whether the effect comes from AI or from relying on ready-made answers. How people use these tools may determine whether they save time while supporting learning, or displace the effort that builds it. And yet, the reported finding about struggling or quitting after assistance ends deserves attention even if it can’t settle the cause. This is more significant than it looks: support that helps someone complete today’s task may not prepare them for the next one unless they also practice working without it.

What the finding means for learners and tool designers

If you’re working in this space, the practical question is not simply whether an AI tool gives a useful answer. It’s whether the person using it can still make progress when that answer is no longer available. For learners, that may mean paying attention to whether a tool clarifies a difficult step or replaces the attempt to solve it. For people building or choosing these systems, the result suggests a reason to consider how assistance is delivered and how users are encouraged to stay involved in the task.

There’s no single design choice that can be recommended from the information provided here, and the study’s reported results don’t establish that every user will respond in the same way. The takeaway is more limited—and more actionable. Ask what happens when the help goes away. A system that makes a difficult task feel easier in the moment may be valuable, but that immediate ease doesn’t answer whether it strengthens a user’s ability or willingness to continue independently.

That’s the tension the poll’s comments circle around: convenience can reduce friction, but effort is also part of how people learn to handle difficult work. The evidence described here raises a question rather than closing the debate. Worth watching.