Original title: How Organizations Use AI: Evidence from ChatGPT
Aaron Chatterji, David Holtz, Neel Rakholia, Prasanna Tambe +1 · August 2026
People talk a lot about artificial intelligence, but how are real workplaces actually using it? Researchers tracked this by connecting ChatGPT Enterprise account records with employee job roles, task classifications, and corporate financial data through March 2026. The system works by analyzing message logs and workplace information while protecting user privacy. Overall, the researchers found that workplace tool use is spreading quickly as new companies join and current users rely on it more heavily. Among American public corporations, adoption is highest in bigger, more valuable companies that invest heavily in research, sales, and administration. Inside these firms, people across many departments and ranks use the system, though early-career employees rely on it with especially high intensity for writing, technical tasks, communication, and summarizing information.
Imagine a company installs a new coffee machine. Within weeks, everyone uses it, but the machine only tracks total cups. The boss wants to know which roles drink the most, so she adds a log for each employee's job title and drink choice. In this paper, researchers use ChatGPT's account data to see which workers—by role and seniority—send the most messages, and for what kinds of tasks like writing or coding. They find that early-career employees are the heaviest users, similar to junior staff grabbing coffee for meetings.
For business leaders and policymakers, this study provides hard data on how generative AI is entering the workplace. The findings show that adoption is uneven—larger, R&D-heavy firms lead, while early-career workers are the most intensive users. This matters because it reveals where AI is actually making an impact and where gaps may persist, helping guide investment, training, and regulation.
The abstract doesn't explicitly state limitations, but the analysis is limited to organizations using ChatGPT Enterprise and may not represent all AI use. The data covers through March 2026, so long-term trends beyond that are not captured. The study documents patterns but does not establish causal effects on productivity or worker outcomes.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
How Organizations Use AI: Evidence from ChatGPT — Aaron Chatterji, David Holtz, Neel Rakholia et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.