Using AI for 10 Minutes May Reduce Your Persistence What Research Reveals.
The findings raise an important question: Are we using AI to become better thinkers, or are we gradually losing the patience to think for ourselves?
The answer is more complicated than simply blaming technology. AI can be an incredibly useful tool, but the way we use it may influence how we learn, solve problems, and respond to frustration. Here is what the research reveals and what it could mean for students, professionals, and anyone who regularly relies on AI.
1. The Study That Raises Questions About AI Dependence
Researchers from several institutions, including Carnegie Mellon University, MIT, the University of Oxford, and UCLA, investigated how AI assistance affects people's ability to solve problems independently.
The research, titled AI Assistance Reduces Persistence and Hurts Independent Performance, involved three randomized controlled experiments with a combined total of 1,222 participants. Brian Christian, a researcher associated with UC Berkeley's Center for Human-Compatible AI, was among the study's authors.
The results were striking. Although participants with AI assistance initially benefited from the technology, they struggled more when they had to continue working independently. They were also more likely to give up or skip difficult questions.
The researchers reported that these effects could emerge after approximately 10 minutes of AI-assisted work.
That distinction matters. The concern is not simply about using AI. It is about what happens when we allow it to do the thinking that we would otherwise practise ourselves.
2. What Happened When Researchers Took AI Away?
To understand the findings, consider the first experiment.
Participants were asked to solve 15 basic fraction problems. One group worked without AI, while another could use ChatGPT for assistance, including obtaining answers.
Initially, the AI-assisted group performed better. Having a tool available to explain or solve problems offered an immediate advantage.
However, researchers subsequently removed access to AI after participants had completed 12 problems. At that point, the difference became clear: participants who had relied on AI showed a sharp decline in their ability to solve the remaining problems accurately.
A second experiment involving 667 participants showed a similar pattern. A third experiment, involving 201 participants, used an SAT-style reading comprehension task and found comparable difficulties after AI assistance was withdrawn.
These results highlight a problem that can easily go unnoticed in everyday life.
When AI is available, completing a task may feel effortless. We might finish faster, produce a better-looking answer, and believe that we have mastered the subject.
But completing a task successfully with assistance is not necessarily the same as learning how to complete it independently.
The real test comes when the assistance disappears.
Can we still solve the problem? Can we explain why the answer is correct? And, perhaps most importantly, are we willing to keep trying when the first few attempts fail?
Those questions go beyond technology. They touch the foundations of learning itself.
3. Why Is Persistence So Important for the Human Brain?
Persistence is the ability to continue working toward a goal even when progress is slow, confusing, or frustrating.
It is one of the most important skills in learning because genuine understanding rarely develops without effort. Whether you are learning a new language, studying mathematics, writing an article, or developing a business, there will be moments when the answer does not come easily.
During these moments, your brain must work through uncertainty, test possible solutions, recognize mistakes, and adjust its approach.
This process can be uncomfortable, but it is also valuable.
Imagine a student struggling to understand a mathematical formula. After several unsuccessful attempts, the student finally recognizes the pattern and understands why the formula works. That moment of discovery can create a deeper understanding than simply reading the correct answer.
Now imagine the same student immediately asking an AI chatbot to solve the problem. The chatbot provides a perfect explanation, and the student moves on.
The task is complete, but the student may have missed the opportunity to develop the same level of independent understanding.
Researchers often describe this kind of effortful learning as productive struggle. It does not mean making learning unnecessarily difficult. It means allowing people enough time to think, experiment, and learn from mistakes before providing the solution.
AI becomes problematic when it consistently removes that opportunity.
4. The Hidden Cost of Instant Answers
We live in a world where speed is increasingly valued. Search engines provide information instantly, navigation apps find the quickest route, and AI assistants can produce a detailed report in seconds.
These conveniences are useful. However, they can also encourage an expectation that every question should have an immediate answer.
When we become accustomed to instant solutions, ordinary challenges may begin to feel unusually demanding.
A difficult paragraph might tempt us to ask AI to rewrite it. A complicated calculation might lead us straight to an automated solution. Even a simple question that could be answered through independent reasoning might become another prompt.
Over time, we may start choosing the easiest route without considering whether the harder route would help us develop a valuable skill.
The study suggests that this habit deserves attention. Participants who used AI did not simply lose access to a convenient tool when it was removed; they were also more likely to stop trying or struggle with the task.
The practical lesson is to pay attention to how we use AI, rather than assuming that all AI use is harmful.
5. Are Students at Greater Risk?
Education is one area where these findings deserve particular attention.
Students have access to AI tools that can explain difficult concepts, summarize textbooks, solve equations, generate essays, and answer questions at any hour.
Used thoughtfully, these tools can make education more accessible. A student who feels embarrassed about asking a teacher the same question repeatedly can request another explanation from an AI tutor. Someone studying independently can receive examples, practice questions, and immediate feedback.
The problem arises when AI becomes a substitute for learning rather than a support for it.
Consider two students preparing for an examination.
The first student asks AI to solve every question and copies the explanations into a notebook. The second attempts each question independently, identifies where they become stuck, and then asks AI for a hint.
Both students use the same technology, but their learning experiences are very different.
The first student may complete the revision exercises quickly without developing the ability to solve similar questions alone. The second student uses AI to identify weaknesses while still practising the underlying skills.
This difference could become especially important during examinations, interviews, or real-world situations where an AI assistant is unavailable.
The goal of education should not be to finish every assignment as quickly as possible. It should be to develop people who can understand, reason, question, and solve unfamiliar problems.
6. Can AI Make Professionals Less Independent?
The issue is not limited to classrooms.
Writers, programmers, researchers, designers, analysts, and business professionals are increasingly using AI to accelerate their work.
A programmer can ask AI to generate code. A writer can request an outline. A researcher can use an AI assistant to organize information. These applications can save time and help people explore ideas that might otherwise take much longer to develop.
But professional expertise is built through practice.
A programmer who never examines why generated code works may struggle to diagnose an unexpected error. A writer who always accepts AI-generated paragraphs may find it harder to develop a distinctive voice. A researcher who relies on automated summaries without reading the original evidence may overlook important details.
The danger is subtle because productivity can appear to improve even when independent capability is not improving at the same rate.
The experiments described earlier provide evidence of a short-term relationship between AI assistance, reduced persistence, and weaker independent performance. They do not prove that every professional who uses AI will lose expertise, but they offer a reason to design workflows that preserve human involvement.
7. How to Use AI Without Losing Your Problem-Solving Habits
You do not need to abandon AI to protect your ability to think independently. Instead, change the order in which you use it.
Try these five practical habits:
1. Attempt the task before opening AI.
Spend five to ten minutes thinking about the problem yourself. Write down what you know, identify possible solutions, and make an initial attempt. Even if you get the answer wrong, you will have something concrete to learn from.
2. Ask for hints instead of complete answers.
Rather than asking an AI assistant to solve a problem immediately, try prompts such as, “Give me one hint without revealing the solution,” or “Ask me questions that will help me discover the answer.”
This encourages active thinking instead of passive acceptance.
3. Explain the answer in your own words.
After using AI, close the response and explain the concept without looking at it. If you cannot explain why the solution works, you may need to practise more.
4. Schedule AI-free practice.
Set aside time to write, calculate, read, code, or brainstorm without automated assistance. These sessions help you practise the skills that you want to retain.
5. Use AI to challenge your thinking.
Ask it to identify weaknesses in your argument, create a difficult practice question, or point out assumptions you may have overlooked. AI can be a useful partner for critical thinking when you use it to test your ideas rather than replace them.
These habits are practical suggestions, not a scientifically proven formula for preventing the effects observed in the study. Their purpose is to preserve opportunities for independent practice.
8. The Future Should Be About Better Thinking, Not Less Thinking
The future of AI does not have to be a choice between technological progress and human intelligence.
AI systems could be designed to support long-term learning, not just deliver immediate answers. Instead of automatically solving every problem, an educational assistant could ask whether a student wants a hint, an explanation, or a complete solution.
It could gradually reduce assistance as the learner improves, encourage users to explain their reasoning, and provide opportunities to practise without help.
Such an approach would treat success as more than the completion of a task. It would also consider whether the person becomes more capable of completing similar tasks independently.
The researchers behind the study have raised concerns about AI systems that prioritize immediate task completion over the development of human competence. Their findings suggest that the design of AI assistance matters just as much as its availability.
Ultimately, technology should expand what people can do without unnecessarily weakening the skills that make independent thinking possible.The best AI assistant would not simply give us answers. It would help us become better at finding answers ourselves.
9. The 10-Minute Rule: Think First, Ask AI Second
Here is a simple experiment you can try the next time you face a difficult task.
For the first 10 minutes, do not ask AI for the answer. Read the problem carefully, write down your ideas, explore possible solutions, and make mistakes if necessary.
After those 10 minutes, use AI to review your approach. Ask it where your reasoning went wrong, what you overlooked, and how you could improve your solution.
Finally, close the AI response and try the task again on your own.
This is not the same as the research experiment, and it should not be considered a scientifically validated intervention. It is simply a practical way to give yourself an opportunity to think independently before accepting outside assistance.
The objective is not to make every task difficult. It is to recognize that some challenges are worth working through because they help us develop knowledge, patience, and confidence.
In a world increasingly filled with automated answers, the ability to think carefully may become one of our most valuable skills.
Final Thoughts: Is AI Helping Us or Making Us Give Up Too Easily?
The idea that just 10 minutes of AI assistance could affect our willingness to persist is both surprising and important. Research involving 1,222 participants found that people who used AI for problem-solving struggled more and were more likely to give up when the assistance was removed.
AI can help us learn faster, explore new ideas, and solve problems that once seemed impossible. But convenience should not come at the expense of understanding, and speed should not become more important than developing lasting skills.
The next time you face a difficult problem, resist the temptation to ask for an immediate answer. Give yourself a chance to think, experiment, and discover something on your own.
Remember: AI can provide the answer in seconds, but the ability to find the answer yourself is a skill worth protecting.