Counting on AI Problem Solving Makes Us More Likely to Struggle and Quit, Research Suggests

Using generative AI A few minutes to solve a complex math problem or write an email can be a quick and easy way to get work done. But your critical thinking, creativity and reasoning skills remain unused when AI does all the hard work for you. That may have more impact than you think.

A recent study conducted by researchers at Carnegie Mellon University, the University of Oxford, the Massachusetts Institute of Technology and the University of California, Los Angeles, found that participants who relied on AI to practice a task for just 10 minutes struggled to do the task without it, compared to those who never used AI in it in the first place.

The researchers said AI chatbots can help with “deep thinking tasks” such as reading again brainstormingbut that relying on bots for these tasks may interfere with our ability to perform those tasks ourselves. There is a big difference between having An AI tool solve problems for you and use it for hints and explanations — or not at all.

The research raises ongoing concerns about how using AI affects our ability to perform similar tasks without the help of a machine. It’s one thing to have such a tool ChatGPT or Claude give us guidance or answer some questions that point us to a solution, but it’s another thing to do all our jobs.

Using AI to complete homework before a deadline is not the same as letting it control your time. Overuse of AI, especially in workplaces again schoolsit may make us, as humans, worse at doing things that we should really be able to do ourselves.

Solving problems, with and without AI

The study involved 1,200 US participants on the research platform Prolific and three surveys. The tests tasked participants with solving fraction problems and answering SAT-style reading comprehension questions. Some people are given AI assistant to get help, and be able to use it as they liked, but it was removed after 10 minutes. The study found that participants who used AI for help were more likely to quit problems and did worse than participants who did not.

The negative effects of AI use were observed only among those who relied on it to solve problems, not among those who did not use it during the study. There were similar findings in MIT study last year focused on using AI to write articles.

Grace Liu, a Ph.D. student at Carnegie Mellon University and one of the study’s authors, told CNET via email that only short-term effects have been measured, and we don’t yet have a full picture of long-term effects. “How significant and long-term this effect is requires further research,” Liu said. Even though AI can be used differently in all use cases, research does not examine how AI tools are used.

The researchers also found that when participants who used AI did not know how to use it, they did not feel confident in their problem-solving skills.

“People not only get worse at tasks, but they also stop trying,” the researchers wrote. “If such effects accumulate over months and years of using AI, we could end up creating a generation of students who have lost the ability to struggle effectively without the support of technology.”

Is AI completely different from a calculator?

The research makes me wonder if using AI is the same as other problem solving methods or shortcuts, such as using a calculator to solve a math problem. Another difference is that generative AI can be used for almost anything, such as personal decisions, planning and research with follow-up questions.

“Both of these conditions are very similar in that they allow people to perform mental tasks,” Liu said. “We believe it is very important to study cognitive abstraction in AI because AI can be used everywhere in all cognitive tasks, while previous tools are task-specific.”

Should you let AI do the work for you?

The researchers say the findings raise questions about the effects of our persistence and thinking when using AI on a daily basis.

“We warn that if these effects accumulate with continued use of AI, current AI systems — designed to help only in the short term — risk destroying the very human skills they are intended to support,” the researchers added.

Liu recommends caution when using AI tools.

“Our results suggest that we should be more serious about how and when AI assistance is used — especially in learning situations,” he said. “It’s not a reason to avoid AI, but it is a reason to design and use these tools carefully.”