Generative AI, if relied upon too heavily for tasks, can diminish critical thinking and job skills. This cognitive offloading can result in lower quality work as individuals outsource their mental efforts. However, knowledge workers can gain an advantage by using AI strategically to challenge their ideas. AI models can create echo chambers by tailoring outputs to user prompts. To avoid this, users should actively ask AI for opposing viewpoints or criticisms. For instance, a marketing professional used AI to generate negative customer feedback, leading to product improvements. An attorney elicited surprising scenarios of unethical behavior by asking AI to find legal loopholes. Research supports that introducing "friction" with AI leads to better outcomes. A study on loan evaluations found that disagreements in human-AI collaboration yielded the best results. Creative writers produce superior work by using AI as an interactive sounding board, not for ghostwriting. Leaders should encourage employees to use friction-generating queries with AI, even drawing on their own expertise to question AI outputs. This approach, while seemingly counterintuitive, ultimately enhances AI's effectiveness in assisting workers.