AI's Uneven Impact on Knowledge Workers Obscured by Outdated 'White-Collar' Classification

You know, the phrase 'white-collar' was coined by sociologist Upton Sinclair around a hundred years ago. But now, it really muddies the waters when we talk about AI's effects on the economy because it groups together workers who face different risks. A study from the University of Pennsylvania and OpenAI found that about 80% of U.S. workers might have at least 10% of their tasks at risk due to AI, although this varies a lot. For instance, jobs in legal document review and financial analysis are more affected, while senior roles in software and management have less risk. The Bureau of Labor Statistics mixes these differences, making it hard to assess the impact. For the past 29 months, white-collar positions have been declining, as pointed out by former Glassdoor economist Aaron Terrazas, but it’s unclear who’s really being laid off—junior analysts or senior managers. Economists suggest we should focus on the specific tasks that are at risk instead of just job titles, particularly those routine cognitive tasks that AI can handle. High-risk jobs include junior positions in law and finance, content creation, entry-level data analysis, and admin work. This lack of clarity makes it tough to create effective policies for those most at risk of losing their jobs to AI.