When AI Does Your Job Faster

49% of employees secretly use AI at work. The other half are in denial. Both responses are destroying your transformation efforts.

Sep 11, 2025

Hi there,

Even though I’ve been writing for a while about the constant capability increases of AI models, the acceleration keeps surprising even tech optimists. But the most interesting pattern isn't the technology. It's how professionals react when machines do their work faster, better and cheaper.

The response is remarkably binary. When Goldman Sachs estimated that AI could automate much of legal work, lawyers split into two camps: early adopters and resisters. Translation tells an even starker divide. Research shows machine translation cost the industry about 28,000 jobs between 2010 and 2023. People either deny it's happening or fall into despair. There's no middle ground.

This pattern paralyses organisations. Nearly half of all workers now hide their AI use, with 49 percent of employees secretly using AI at work according to recent surveys. When your best people are either denying reality or hiding their adaptation, nothing moves forward.

On the other side of the issue is a persistent cohort of denialists, refusing to believe that this major societal change is happening, and jumping on any-and-all mistake or error an AI model makes to justify their continued dismissal and denial.

Your Job Becomes Your Identity

To understand why smart people fall into this trap, consider what happens when your expertise defines who you are. Professional identity is more than just a job. It's years of training, specialised knowledge, and a sense of self that goes way beyond your current role.

When AI threatens that identity, it's not just about losing work. It's about losing yourself. Herminia Ibarra describes it perfectly: professionals get stuck "oscillating between holding on to the past and embracing the future." Rebuilding your entire professional identity means testing new versions of yourself, finding new networks, and making sense of a completely different reality. It's exhausting.

The grief is real. Watching your expertise moat disappear genuinely hurts. The knowledge that once guaranteed your value no longer provides the same protection. Yet most companies treat AI adoption like it's just about learning new software, not processing profound change.

What Collapse Looks Like

Translation shows what happens when a profession doesn't adapt. For every one percent increase in machine translation use, translator employment growth dropped by 0.7 points. The industry lost about 28,000 potential jobs between 2010 and 2023. Translators tried to reinvent themselves as "post-editors" who would polish machine output. Many found that fixing machine translation takes longer than translating from scratch. Catching subtle errors turned out to be harder than doing original work.

Radiology took a different path. Studies show AI helps radiologists spot things they might miss. One radiologist describes his success: "With a high volume of high acuity cases, AI helps us manage our caseload efficiently. Ninety to ninety-five percent of impressions I don't have to edit."

The key difference? Radiologists became orchestrators rather than operators. They integrate data from multiple sources, supervise AI systems, and focus on complex cases that need human judgment. Translation tried to save the old model. Radiology built a new one - although many holdouts still resist AI despite the evidence!

The Orchestrator Advantage

Moving from doing the work to directing the work isn't just wordplay. It's a fundamental shift with real financial benefits. Some jobs requiring AI skills pay 23 to 28 percent more according to labour market data. Some markets show even bigger premiums.

This new role focuses on directing, not doing. Orchestrators check AI's work, handle ambiguous situations, build trust with stakeholders, and make ethical calls that AI can't. They connect dots across different areas and create meaning for other humans. The work changes, but your expertise and agency remain valuable.

The successful radiologist doesn't just use AI. She runs it within a broader system. AI's accuracy doesn't reduce his value. His ability to direct that accuracy toward better patient outcomes increases it. That's the model: humans provide judgment, context, and accountability while AI handles volume and patterns.

The challenge is that the duration of this skill premium, even it even persists, will be much shorter than many expect because of the rapid progression of AI model capability.

Share Brennan McDonald's Newsletter

Time Is Running Out

The window for adaptation is closing fast. Research projects more than 12 million Americans will need to switch careers by 2030. Globally, between 75 and 375 million workers may need to change job categories entirely. The old path from junior to senior roles is breaking down as automation changes entry-level work.

Companies investing billions in retraining see results. Successful programmes report 82 percent job placement rates within months, with most participants still employed after a year. But success requires acknowledging the grief when expertise no longer protects you the way it once did. Find where human judgment still matters. Create safe spaces to experiment with AI rather than compete against it. Celebrate people who orchestrate, not just execute.

As AI capabilities improve, fewer professionals have enough of an expertise edge to feel secure. The moat that once protected knowledge workers is shrinking. Not because expertise has no value, but because AI can now match or exceed human performance in areas we thought were safe. The professionals who thrive won't be those with the deepest moats, but those who learn to bridge them.

If your team questions their worth in an AI world, you're facing the right problem. When people get stuck between denial and despair, organisations freeze or experience project failure just when they need to move faster. I help business owners and leaders unfreeze AI transformations by addressing the change management challenges that stop technical progress cold.

PS: I’m going to be writing more articles on LinkedIn each week, the first was called “Your 10-Year Expert Just Became a Beginner”. Let me know what you think.

Brennan McDonald's Newsletter