As artificial intelligence reshapes work across industries, a paradox is emerging among those who understand the technology best: many AI professionals deliberately avoid automating their most basic tasks, choosing pen and paper over cutting-edge tools they help create.
Stella Dong, a machine-learning engineer and co-founder of AI startup Reinsurance Analytics, writes her own emails rather than letting algorithms draft them. “I don’t trust AI to draft by itself,” says Dong, who also works at a healthcare-technology company. She sometimes uses Microsoft Copilot to revise messages but insists on writing initial drafts herself. Dong also manually inputs calendar entries, believing she’s more likely to remember appointments she schedules personally.
The selective approach from AI insiders comes as a new McKinsey Global Institute report, released November 24, estimates existing technology could automate 57% of Americans’ work hours. Yet the consulting firm cautions this figure reflects technical potential rather than inevitable job loss.
When Expertise Means Knowing What Not to Automate
Ziyi Liu, an AI research intern at Microsoft and doctoral student at the University of Southern California, takes typewritten notes during meetings despite having access to software that automatically transcribes and summarizes sessions. “I want to be that kind of person that keeps their work very structured and clear,” Liu says. “I don’t want to look at a transcript; I just want to do it myself. It makes me feel like I’m in control”.
Ryan Bearden, a marketing consultant who trains business teams on AI tools, takes the analog approach further by handwriting notes in a Moleskine notebook. Bearden says the practice helps him commit information to memory while signaling undivided attention to colleagues. His client presentations begin as physical storyboards sketched on printer paper before being refined with tools like Claude or ChatGPT.
“There’s a tendency for folks to jump at a solution when there may not even be a problem that exists,” Bearden says. “AI is a very powerful tool—it’s a hammer and that doesn’t mean everything is a nail”.
The Human Element in an Automated Future
Lareina Yee, a senior partner and director at McKinsey Global Institute, suggests organizations may deliberately hold back automation to preserve essential skills. “As we redesign work and jobs, you might actually choose not to maximize how much an AI agent or robot does,” Yee says. Companies might maintain manual tasks to train junior employees in foundational skills and ensure workers can verify AI outputs.
Research supports the cognitive benefits these professionals intuitively recognize. Studies published in journals including Psychological Science show handwriting notes leads to better retention and understanding compared to typing, as the physical act engages deeper cognitive processing.
The selective automation practiced by AI experts raises questions about how businesses should approach the technology. As an EY survey of 15,000 employees across 29 countries found, while 88% use AI in daily work, most limit usage to basic applications like search and document summarization, with only 5% using it in transformative ways. The gap between AI adoption and effective implementation suggests that knowing when not to automate may be as important as knowing how to use the tools themselves.