Just 10 minutes of AI use can weaken problem-solving: Study
What's the story
A new study from researchers at Carnegie Mellon, MIT, Oxford, and UCLA has revealed that using AI chatbots for even a short period of time can have a negative impact on our cognitive abilities. The research found that just 10 minutes of interaction with an AI assistant could impair problem-solving skills. The findings were based on three experiments involving several hundred participants who were asked to solve various problems through an online platform.
Cognitive effects
AI's double-edged sword
The study found that when participants were given access to an AI assistant that could solve problems on its own, they were more likely to give up or make mistakes when the AI was suddenly taken away. "AI can clearly help people perform better in the moment, and that can be valuable. But we should be more careful about what kind of help AI provides, and when," said Michiel Bakker, an assistant professor at MIT.
Caution advised
Rethinking AI tools
Bakker added, "The takeaway is not that we should ban AI in education or workplaces." He finds the study's results particularly concerning, noting that a person's persistence in problem-solving is not only vital for skill acquisition but also a key predictor of their long-term learning capacity. Bakker suggested that it may be necessary to rethink how AI tools work so they prioritize a person's learning over solving a problem for them.
Behavioral impact
Addressing the issue
AI companies are already considering the subtle effects their models can have on users. OpenAI has toned down the sycophancy of some models in newer releases of GPT. However, over-reliance on AI could be problematic when these tools don't behave as expected. Agentic AI systems, which do complex chores independently and can introduce odd errors, are particularly unpredictable.