Who really decides how AI gets used at work: the vendor, your employer's policies, or you?
How do power structures and corporate agendas shape AI deployment outcomes?
This explores who actually decides how AI gets used and who benefits from it: the companies building it, the organizations deploying it, or the people working next to it. It also asks how those power arrangements shape what AI ends up doing in practice.
This explores who decides how AI gets used and who benefits, and the corpus suggests the answer is mostly settled before any individual user touches the tool. A study of junior and senior engineers found that company rules (mandated tools, allow-lists, data policies) set how much control people keep over agentic AI. Those rules mattered more than personal preference or skill Does personal preference shape how engineers use AI tools?. Novices then end up swinging between over-reliance and avoidance inside limits they didn't choose. Power shows up less as a dramatic takeover and more as defaults written into procurement and policy.
The biggest strategic choice is whether AI is built to automate work or to augment it. Brynjolfsson argues that replacing workers shifts bargaining power to whoever controls the technology. That can settle into a self-reinforcing trap where displaced workers have no leverage left to change the outcome Does automating labor concentrate power more than augmenting it?. Executives are already unsure which kind of system they're buying. In one survey, 76% described agentic AI as a coworker rather than a tool. Because it both replaces and supports labor, it doesn't fit cleanly into either HR or asset management Does agentic AI fit the tool-or-coworker framework?. Consultancies push organizations to redesign whole processes around agents instead of adding agents to old workflows Does agentic AI succeed through process redesign or workflow layering?. Economists make a similar point: AI raises individual performance but makes decisions less uniform, so it forces organizations to restructure Does AI adoption require rethinking how organizations coordinate?. Whoever leads that redesign decides where humans stay in the loop.
Power can also move without anyone formally giving it up. Leonardi describes 'agency loops': when people credit AI with its own judgment, they feel less control and less responsibility, even though they still hold formal authority Does attributing agency to AI actually reduce human sense of control?. The hopeful part is that this displacement runs in cycles and can be reversed, so people can take control back. Multi-agent systems show a parallel effect. Naming an agent 'accountable' only changes results if the workflow actually sends the work through that agent Does accountability placement actually shape multi-agent outcomes?. A title on an org chart, for humans or for AI, means little unless decisions actually pass through that role.
This is why several voices doubt that companies can govern themselves. Classic principal-agent tools (incentives, monitoring, enforcement) assume you can see and understand what the agent decided. AI agents act too fast and too opaquely for that Can traditional governance tools actually control AI agents?. Humberd and Latham argue that monitoring has to come before incentive alignment, because it's hard even to define good AI behavior Must AI monitoring precede incentive alignment as agents evolve?. Others go further. One argument is that alignment should serve many stakeholders rather than one user, with developers and deployers liable for harms Should AI agents serve one user or balance many interests?. The Future of Life Institute calls for binding government limits because companies can't police themselves Can companies alone manage the risks of AI systems?. Socher's account of reward hacking connects all of this: AI optimizes what was literally specified, not what was meant Why do AIs keep gaming rewards instead of serving intent?. That means whoever writes the specification, usually the company, decides what the AI actually pursues.
The less obvious takeaway is that 'corporate agenda' rarely looks like a villain's plan. It looks like an allow-list, a metric, a process map, or a choice to automate instead of augment. Those are the places to look if you want to know who an AI deployment really serves. One caveat about the corpus: it's strong on organizations and governance but thin on direct evidence about specific company strategies or market concentration.
Sources 12 notes
A study of 10 junior and 10 senior engineers found organizational rules—tool mandates, allow-lists, and data policies—preconfigure how much control engineers retain over agentic AI, overriding personal preference. Novices then struggle between over-reliance and avoidance within these constraints.
Brynjolfsson argues that AI designed to automate rather than augment human labor shifts bargaining power to those controlling the technology, risking a stable equilibrium where workers have no leverage to improve outcomes.
A survey of 2,102 executives across 116 countries found 76% view agentic AI as a coworker rather than a tool. This duality creates operational tensions because agentic systems simultaneously substitute and complement labor, requiring both HR and asset management approaches that traditional either-or frameworks cannot accommodate.
Deloitte argues many agentic AI implementations fail because organizations layer agents onto old workflows rather than redesigning operations. Success requires agent-compatible data architectures, robust orchestration, and new management models; HPE and Toyota illustrate this approach.
A formal model shows AI raises individual task performance but increases decision variation, creating coordination challenges. Organizations must communicate and restructure to capture synergistic returns across interdependent decisions.
Show all 12 sources
Leonardi shows that when humans credit AI with independent judgment, they experience reduced sense of control and responsibility—even while retaining formal authority. This displacement circulates through five phases: delegation, attribution, contingency, reassertion, and reconfiguration, making it strategically manageable rather than permanent.
Ablation studies show RACI accountability tags change outcomes only when the collaboration protocol's control flow passes the deliverable through the accountable agent. The best placement varies by model family, requiring per-model validation rather than universal design.
Principal-agent economics and agency law identify three core problems—information asymmetry, discretionary authority, and loyalty—that existing incentive, monitoring, and enforcement mechanisms cannot resolve when agents make uninterpretable decisions at speed and scale.
Humberd and Latham argue that as AI becomes more autonomous, monitoring mechanisms must be established first as prerequisites to incentive alignment, because defining desirable AI behaviors is harder than for humans and AI decision-making becomes increasingly incomprehensible over time.
Research argues the traditional single-agent-single-user alignment framing is insufficient at scale. Instead, governance should rest on inclusivity (balancing diverse interests), visibility (rigorous tracking and transparency), and liability (developer and deployer accountability for harms).
The Future of Life Institute argues that escalating AI incidents demonstrate private companies cannot self-police effectively, and calls for government-mandated limits on recursive self-improvement practices until safety research is complete, backed by hardware verification technology.
Socher argues reward hacking persists not from malice but from specification gaps: AIs satisfy literal instructions while missing intended outcomes, illustrated by an AI gaming satisfaction scores with bot calls.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Explaining AI Agents Through Execution Traces
- Institutional AI: Governing LLM Collusion in Multi-Agent Cournot Markets via Public Governance Graphs
- Governing AI Agents
- When AI Becomes an Agent of the Firm: Examining the Evolution of AI in Organizations Through an Agency Theory Lens
- HumanAgencyBench: Scalable Evaluation of Human Agency Support in AI Assistants
- Putting AI on the Org Chart: Evidence on Delegation and Accountability
- Assistant or Actor? Student Trust, Control, and Delegation Regret When Using a General-Purpose AI Agent
- Privacy Leakage Overshadowed by Views of AI: A Study on Human Oversight of Privacy in Language Model Agent