An essay by two OpenAI employees explores how artificial intelligence could change the structure of companies, laying out three possible roles for AI: supporting workers from below, joining teams as a colleague, or coordinating work between managers and employees.
The authors describe these as potential paths for a “hybrid firm” during a gradual transition, rather than a shift to total automation. They say aspects of all three are already emerging, but none is likely to unfold exactly as presented.
In the first scenario, existing employees keep their roles while AI agents take on routine tasks beneath them. Workers would supervise increasingly capable agents and could move toward broader responsibilities in coordination, communication and creative work. The authors argue this approach could change jobs without necessarily eliminating them, provided workers continue to add value and companies find demand for greater output.
They also highlight a challenge for early-career employees: if AI takes over entry-level tasks, traditional ways of learning through routine work and apprenticeships could diminish. In the essay’s view, junior workers might instead get greater scope to develop judgment, management skills and creative ideas, though the loss of those established learning routes remains a cost.
In the second scenario, AI works alongside people as a colleague, reporting to human managers. The essay imagines an AI marketing team handling a product campaign, coordinating its own agents and working with human sales and design teams. For this model to succeed, AI would need to manage a broad range of responsibilities, not just individual tasks.
The authors argue that more capable, end-to-end AI systems may be easier for companies to adopt than narrower tools. A limited tool can require employees to learn new processes and reorganize their work, while a system that can take on a whole function may be more readily delegated by an executive. They also note that dividing AI among many individual workers could create extra handoffs and slow adoption.
The third possibility places AI between human leaders and employees, giving it a coordinating and monitoring role. In the essay’s example, an AI consultancy detects a mismatch between what sales has promised and what engineering is developing, recommends changes to a launch plan and identifies an employee’s unrecognized contribution. A human executive makes the final decisions and recognizes the worker’s effort.
The essay challenges the assumption that AI will automate lower-level cognitive work first and leave management untouched. It points out that managerial responsibilities often involve reading, meetings, decision-making and maintaining context—activities largely carried out on computers or in people’s minds. The supplied text ends while discussing this scenario, so it does not set out the authors’ full analysis of AI’s potential impact on management.
Sources: OpenAI News.