The Moment Students Lose Their Thinking
A university student opens ChatGPT minutes after receiving a creative problem-solving assignment. Within seconds, the AI generates several polished solutions. The student picks one, makes a few small edits, and submits it. On paper, it looks like sophisticated work. But something vital has been lost: the student's independent idea generation. The AI completed most of the cognitive work before the learning process had even begun.
This scenario plays out in classrooms everywhere. While generative AI has enormous potential to support learning, the question is no longer whether students should use it. It's how AI can support learning without replacing human reasoning.
A 2025 study by Guo and colleagues, Student–AI Creative Problem-Solving: The Role of Human Agency, takes this question seriously — and finds that simply telling students to "think first" before using AI isn't enough. Agency must be sustained throughout every interaction, not just established at the start.
What Human Agency Actually Means
In the context of AI-supported learning, agency means students' ability to intentionally direct their own learning by contributing to and maintaining ownership over the problem-solving process. It's not about whether students use AI at all. It's about whether they remain active contributors throughout the task, or whether they gradually become passive recipients of AI-generated content.
The distinction matters. When students treat AI as an answer generator rather than a thinking partner, they outsource the learning process. And research shows this happens faster than most educators expect.
The First Experiment: Why "Thinking First" Isn't Enough
Guo et al.'s study is particularly valuable because it doesn't simply compare AI users with non-AI users. Instead, it investigates how different patterns of collaboration shape students' creative thinking. Across two quantitative experiments involving 346 university students, participants completed the same creative task: developing innovative ways to improve a toy bunny to increase sales.
In the first experiment, 170 participants were divided into three conditions. One group generated their own ideas before consulting ChatGPT (pre-SHT). Another group consulted AI first and then added their own ideas afterwards (post-SHT). A control group relied on AI throughout without any instruction to engage in independent thinking (without-SHT).
The results were revealing. Students who generated their own ideas before using AI reported greater initiative, stronger ownership of their work, and invested more mental effort than those who immediately turned to AI. This pre-SHT group also produced more cognitively sophisticated prompts that reflected higher-order thinking rather than simply repeating information from the task.
Here's the catch, though: this advantage gradually disappeared as students continued collaborating with AI. By the third prompt, many participants had begun following the AI's suggestions instead of directing the collaboration themselves.
Simply delaying AI use encouraged independent thinking initially, but it was insufficient to sustain human agency throughout the task. That's a sobering finding, and one that educators need to hear.
The Solution: Deep Idea Integration
Recognizing this limitation, Guo et al. redesigned the collaboration in a second experiment involving 176 students. Instead of asking participants to think independently only at the beginning, they introduced a Deep Idea Integration (DII) condition, which required students to integrate their own ideas into every interaction with ChatGPT.
Rather than asking broad questions and accepting AI-generated solutions, participants first generated an original idea before using AI to refine, challenge, or extend it. For example, instead of asking ChatGPT how to improve the toy bunny, a student who had already proposed making the toy glow safely in the dark might ask which child-safe materials could achieve this or whether similar products already existed.
In this condition, students — not the AI — determined the direction of the collaboration.
The learning benefits of the DII constraint were striking. Students maintained higher levels of perceived agency throughout the task. They produced prompts that remained cognitively sophisticated across all stages of the collaboration. And they generated final solutions that were significantly more novel and useful than those in other conditions.
Their final design solutions were also less similar to one another, suggesting that students were developing genuinely original ideas rather than converging on the AI's preferred solutions. Importantly, participants also reported greater perceived cognitive improvement, indicating that the collaboration enhanced their own learning experience rather than simply improving output quality.
What Other Research Tells Us
Similar findings appear elsewhere in the literature. Xia et al. (2026) found that while GenAI-supported group work assisted learning and metacognitive engagement, groups working without GenAI in their experimental study produced more original creative thinking solutions.
Students also report these consequences directly. Perifanou and Economides (2025) found that students valued GenAI most when it was used to clarify difficult concepts rather than replace independent learning.
The broader pattern is unmistakable: when students use AI without maintaining their own intellectual contribution at each step, the cognitive benefits erode. The Education Endowment Foundation's ongoing £2.5 million investigation into how GenAI affects student cognition is examining exactly this question at scale, with particular attention to cognitive offloading in adolescents. Read about the EEF's GenAI cognition study.
This dynamic mirrors what researchers call the "AI Dependency Paradox" — a phenomenon where initial performance gains from AI assistance give way to long-term degradation of independent skills. A recent MIT Media Lab study found that users who habituated to AI-assisted verification became significantly worse at detecting misinformation when the tool was removed. Learn about the AI dependency paradox. The same pattern appears in education: students who rely on AI for homework see short-term grade boosts but suffer long-term declines in independent exam performance, a phenomenon researchers term "cognitive foreclosure." Explore the cognitive debt of AI schoolwork.
Together, these findings reinforce Guo et al.'s central conclusion: AI supports learning most effectively when educational design requires students to remain active contributors rather than passive recipients throughout the learning process.
Designing for Agency in the Classroom
Guo and colleagues measured agency primarily through students' self-reported perceptions, which may not capture the full complexity of how agency operates and develops during learning. Despite this limitation, the findings align with broader educational theories that position agency as central to effective learning.
The implication for education is clear: don't suppress AI use. Design interactions that preserve students' intellectual independence, curiosity, engagement, and understanding of how AI can influence learning.
Generative AI can act as a powerful learning partner when students use it to refine and expand their own ideas rather than outsource the learning process. As AI becomes increasingly embedded within higher education, educators will need to develop assessment and teaching strategies that reward the learning process as much as the final product.
Encouraging students to critically question and build upon AI-generated suggestions will help ensure that AI strengthens, rather than diminishes, the intellectual skills and human agency of students.
The Bottom Line
The research is clear: agency in AI collaboration must be sustained throughout a task, not just established at the start. Students who thought first before using AI still drifted toward following AI by the third prompt. But requiring an original idea at every prompt kept students directing the AI, not the other way around.
The lesson for educators is practical and actionable. If you want students to think independently while using AI, you need to design the interaction so that independent thinking isn't a one-time event at the beginning. It needs to be a continuous requirement — renewed at every prompt, every step, every stage of the collaboration.
Because the alternative — letting AI do the thinking — is already happening. And it's happening faster than most of us realize.
Source: Michael Hogan, Ph.D., lecturer in psychology at the National University of Ireland, Galway. Co-authored by Ava Gilmartin and Paul Surlis. Psychology Today article, July 26, 2026.
