Botsitting, botshitting, and the hidden human labor of AI at work (www.glean.com)

A recent analysis highlights the complex relationship between AI tools and human labor in workplaces, emphasizing that organizations often mistakenly try to enhance AI transformations through mere increases in tool usage. This "addition sickness," as noted by Stanford Professor Bob Sutton, leads to tokenmaxxing, where companies focus on maximizing AI tokens rather than improving productivity. Instead, successful organizations prioritize defining the context in which AI operates and investing in the human infrastructure necessary for effective AI utilization. Workers who excel with AI, termed "high AI achievers," distinguish themselves by deliberately limiting AI's role to non-core tasks, thereby maintaining their cognitive skills and judgment. These high achievers engage more in what is termed "botsitting," where they oversee AI outputs and refine prompts, leading to higher-quality work and learning opportunities. The study illustrates that effective AI interaction involves not just using tools but understanding when not to use them at all. Interestingly, many workers engage in constructive deviance, bypassing organizational policies to utilize unapproved tools when sanctioned options fall short. Organizations are encouraged to recognize these workarounds as vital feedback for reassessing their AI strategies, aiming to streamline AI use while fostering innovative work environments.