OpenAI researchers envision three organizational models for AI-augmented firms
Two young OpenAI managers outline how AI labor might restructure corporate hierarchies—from universal promotion to hybrid human-AI coordination.
Last verified:
AI as Subordinate Labor: The “Beneath Us” Scenario
According to OpenAI’s blog essay, two researchers at the company—both managing teams at age 24—propose that AI labor could function as a new organizational basement. In this model, every employee gains a promotion: individual contributors become managers overseeing AI “deputies” that handle routine work. The researchers describe these AI reports as tireless performers of “grunt work,” capable of learning managerial preferences over time, though imperfect. Math Olympiad gold medalists might advise junior staff, while AI systems execute foundational tasks. The outcome is organizational leveling—everyone moves up one rung.
AI as Peer Colleague: The “Beside Us” Model
A second pathway positions AI as a lateral entrant to the org chart. According to the OpenAI essay, AI systems would report to existing human managers and executives, starting with discrete tasks before absorbing broader functional responsibilities. This scenario resembles hiring a new generalist colleague rather than deploying a subordinate workforce. The progression moves incrementally: AI takes “this task and that role” before ascending the hierarchy and reporting to higher-level decision-makers.
AI Between Humans: The Coordination Model
The third and least intuitive scenario places AI in a middle layer—supervising execution while respecting human initiative. According to OpenAI, this “between us” arrangement has humans above (executives setting vision) and humans below (enterprising workers), with AI enforcing decisions and monitoring quality. This model treats AI as coordinator rather than subordinate or peer, translating leadership intent into operational oversight.
Hybrid Adoption as Path Dependence
The researchers emphasize that adoption isn’t uniform across capability levels. According to the blog post, junior employees—already embedded in task-level workflows—may be fastest to productively integrate AI labor. As models mature and perform end-to-end functions, higher-ranking executives can adopt AI more directly. The authors argue that “more capable AI is more adoptable” because labor learns the firm faster than firms learn new technologies; workers can onboard themselves into AI collaboration.
Why This Matters
The OpenAI researchers sidestep the total-automation narrative to focus on structural reorganization during a “transitioning economy.” Their three scenarios have implications for power distribution, compensation inequality, and firm scale. If junior workers command AI subordinates, hierarchies flatten and pay disparity may widen (fewer layers means fewer middle-management jobs). If AI sits between executives and workers, centralized control increases. The timing and sequence of AI integration—ground-level first, executive-level later—will determine which outcome dominates. This framework gives organizations permission to experiment rather than wait for complete AI autonomy, potentially accelerating the reshaping of work itself.
Frequently Asked Questions
Do these OpenAI researchers think AI will replace workers?
No. They focus on 'productive adoption' during a 'transitioning economy,' not total automation. They describe chimeric (hybrid) firms where humans and AI coexist in restructured roles.
Which organizational model do they predict will dominate?
They expect aspects of all three to emerge, but suggest the 'between us' model—with AI coordinating between executives and ground-level workers—may be most practical as AI capabilities broaden.
Why does this matter for workers?
The structure OpenAI researchers describe could reshape pay disparity, firm size, and worker agency. Junior employees might gain managerial power sooner (via AI subordinates), or see authority centralize around AI coordinators.