OpenAI Researchers Propose Three Scenarios for AI Integration in Corporate Structures

Details

OpenAI researchers Hemanth Asirvatham and Elliott Mokski outline three scenarios for AI integration: Promoting (AI beneath humans), Interleaving (AI beside humans), and Layering (AI between humans).
The core argument is that human value lies in coordination and partial execution of large outputs, not individual task completion; economic AGI must operate at the level of 'people,' not just 'persons.'
AI acting as labor (active initiative) lowers adoption barriers compared to AI as technology (requiring retraining), enabling faster integration into existing workflows.
Management roles may be automated before lower-level tasks because managerial work is less entangled with physical tools, though current models lack evaluation benchmarks for large-scale orchestration.
AI consulting agents can centralize power by reducing information friction between C-suite and employees, while also potentially increasing employee agency by providing high-quality advice to all staff.
Firm size dynamics will shift: AI as a coordination layer makes larger firms less inefficient, while frictionless communication may reduce the need for internalizing supply chains, leading to a mix of small firms and large AI holding companies.
Wage inequality may increase if human judgment and scope become the primary differentiators, as the value of good vs. bad decision-making becomes more pronounced in AI-led firms.

Key point