The Human-AI Inflection: Why Efficiency Is Not the Only Metric

In this look at Joanna Stern’s year of working with AI, we see a change in the workplace: the shift from AI as a tool to AI as a partner. The hidden cost of this change is not that humans are being replaced, but that we are losing the struggle. This friction is what builds expertise and professional value. While companies use AI to become more efficient, the long-term risk is a workforce that lacks the foundational experience needed for complex tasks. For leaders, the competitive advantage comes from resisting the urge to automate the hard parts. By keeping human friction in key workflows, leaders can build a defense that AI-driven competitors, who focus only on speed, cannot copy.

The Efficiency Trap and the Erosion of Expertise

The most common mistake in adopting AI is the attempt to remove every bit of friction. Stern’s research into customer service shows a clear, dangerous pattern. Companies use AI to sort, template, and answer repetitive questions, which lets them grow without hiring more people.

The immediate benefit is clear: costs go down and response times improve. But the long-term effect, as Stern’s former research assistant noted, is a skill gap for the next generation. If junior employees never do the repetitive, foundational work that serves as an apprenticeship, they never learn the pattern recognition needed for high-level problem solving.

I think her response was really good. It was that she was more worried about some of the younger folks coming out of college. She had already had a couple years under her belt working as a reporting assistant and as a reporter, but she was worried how are they going to get that experience versus yeah she is like I do not really love doing some of that work for you, right? Now I can work on some more higher level tasks.

-- Joanna Stern

The First Pass Fallacy

Many assume that using AI for a first pass is always a win. It saves time and sets a baseline. However, the system responds to this shortcut by lowering the quality of the final result. When we rely on AI to organize our thoughts or write our first drafts, we skip the mental work that creates our unique perspective.

Stern’s experience interviewing experts with an AI avatar shows the mediocrity that comes with automated processes. The AI can make a transcript, but it cannot capture the nuance, the redirection, or the human insight that makes a conversation valuable. The danger is that we mistake faster for better. When we let an algorithm handle the synthesis, we are not just saving time; we are outsourcing the thinking that makes our work distinct.

The hardest part for me in that process is taking all of that information and narrowing it down to a story that I think is interesting. AI can do that, I do not know if it can actually do it that well, but AI can do that and I could use AI as a crutch there or maybe as a first pass, but there are some people who will say, okay, AI can do it, I am just gonna have it do it, right?

-- Joanna Stern

The Competitive Advantage of Hard

The most important insight from Stern’s work is that the current AI backlash is not a sign of failure, but a necessary correction. People are realizing that easier often comes at the cost of environmental, economic, and societal health.

In business, the choice to keep a human in the loop, even when an AI could do the task, is a strategic one. It creates a barrier to entry. If your competitors are automating their thinking, they are getting faster, but they are also getting more generic. By choosing to keep the hard parts of your process, you ensure your team stays capable of deep, non-obvious work.

It is supposed to be hard if it was not hard everyone would do it.

-- Joanna Stern (quoting A League of Their Own)

Key Action Items

• Perform an AI Applicability Audit: Every six months, map your team tasks by administrative drudgery versus high-level thinking. Automate the former; protect the latter. (Immediate)
• Implement Human-in-the-Loop Mandates: For critical decisions or creative output, require a human to perform the synthesis before reviewing AI-generated drafts. This prevents the anchoring effect of AI suggestions. (Immediate)
• Design Apprenticeship Pathways: If AI is replacing entry-level tasks, you must build new ways for junior staff to encounter productive struggle. This is a 12-18 month investment to prevent long-term skill loss. (12-18 months)
• Resist the End-of-Day Crutch: Establish a rule that when you are tired, you do not use AI for high-level synthesis. Fatigue-driven automation leads to low-quality, generic output that compounds over time. (Immediate)
• Prioritize Human-First Interactions: In roles involving networking or high-stakes communication, use AI only for scheduling and administrative prep. Never automate the actual interaction or the strategic follow-up. (Ongoing)