Research reveals how AI enables workers to perform tasks beyond their traditional roles

OpenAI's analysis of over 800,000 messages from U.S. ChatGPT users has uncovered significant shifts in how work responsibilities are distributed across organizations. The data shows that 16.8% of work-related messages and 43.5% of occupation-specific messages involve tasks historically performed by other professional categories, a pattern researchers term "task crossover". The implications become clear through practical examples: small business owners independently handle document drafting and contract review, salespeople analyze customer datasets previously managed by analysts, and marketers troubleshoot website issues without developer involvement. This phenomenon indicates that AI fundamentally reshapes not only work methods but also the allocation of responsibilities among workers. The analysis reveals distinct patterns across professions. Designers frequently engage in cross-occupational tasks, comprising 35.2% of their messages, while engineering work appears extensively in communications from non-engineers. Marketing emerges as particularly distinctive, accounting for tasks drawn from multiple domains while simultaneously spreading widely throughout the organization. Organization size plays a significant role in shaping these patterns. In smaller businesses, workers closer to problems tend to assume responsibility rather than delegate, resulting in higher cross-occupational task rates: 18.9% in teams with 2-5 members compared to 16.3% in larger organizations exceeding 100 employees. These usage patterns provide early indicators of occupational transformation that traditional labor statistics will only capture later.