How might ‘arguing’ with AI improve the quality of your work?

Recent research indicates that engaging AI in a critical manner can enhance the quality of work for knowledge workers. Scientists, teachers, lawyers, and business leaders are increasingly relying on generative AI applications for tasks that require imagination, creativity, and problem-solving. However, interacting with AI can sometimes lead to cognitive offloading, where workers rely on AI's confident answers, resulting in low-quality solutions.

This study aims to explore how intentionally using AI to challenge one's ideas can improve performance. Researchers interviewed 45 knowledge workers at a US public university, who were using generative AI applications in tasks like generating new ideas and solving complex problems. The study found that intentionally using AI to generate opposing viewpoints can significantly improve performance.

For example, a marketing professional asked AI to create virtual customers who would dislike a product, revealing customer preferences that the professional had not considered. Similarly, an attorney used AI to find legal loopholes and consider unethical company behaviors. The study also highlights the importance of questioning AI outputs and leveraging one's expertise to understand unfamiliar information.

Leaders of organizations are encouraged to foster friction-generating queries of AI models, using prompts that challenge rather than agree with a person's thinking. Offering resources such as AI bootcamps, workshops, and sandboxes can also help workers understand the capabilities and limitations of AI tools.