“Tell me about this recommendation,” a manager says to a junior associate.
The employee looks at the document and thinks for a second. How much of their own thinking went into this and how much was it AI leading the way? How much could they explain on the spot? Was there something wrong or was it an honest question?
This exchange is happening more and more in roles where AI can do serious work. While the public conversation about this tends to focus on entry-level roles, AI has impacted everything from an individual contributor’s detailed project plan to an executive’s board deck.
The problem is that few people can tell the difference between someone who understood the material and used AI to move faster and someone who never understood it at all. The document looks plausible either way.
Job architecture runs on the assumption that a title predicts the work inside it. That assumption is breaking.
How 7,000 Workers Upended Our Understanding of Job Architecture
ADP Research and Stanford’s Digital Economy Lab tracked about 7,000 IT workers from 2019 to 2025 using real payroll data. The researchers looked at 5 million job postings and 9 million payroll records across more than 600 employers. They then mapped the work into 25 standard task categories and ran the wage data against each one. They attempted to control for age, gender and company size to find which specific tasks earned a premium and which didn’t.
Continue Reading: Why Managers Can’t Tell AI Skill From an AI Shortcut