Why do I want to research this?

I’m currently applying for admission to the Doctor of Business Administration (DBA) programme at Curtin University.

My research proposal title is: Who Decides When the System Acts? Decision Rights, Accountability and Operating Model Archetypes for Agentic AI in Large Enterprises

The starting point for me is quite simple.

For most of the history of enterprise technology, systems have helped people make decisions. We could therefore design organisations around a reasonably clear assumption: that a person or group has the authority to decide, that technology informs that decision, and that somebody can ultimately be held accountable.

Agentic AI changes that relationship.

We are now introducing systems that can interpret a goal, plan how to achieve it, choose between alternatives and take actions across enterprise systems, sometimes without a person approving each individual decision.

That creates an organisational problem that I think is at least as important as the technical problem.

• Who is actually deciding?
• Who gave that authority away?
• What is the AI allowed to decide?
• Who defines the information and business meaning on which it acts?
• When must it escalate?
• And who is accountable when the person formally responsible was not the one who actually made or executed the decision?

There is already very good research around parts of this problem.

Weill and Ross explain how organisations allocate technology decision rights. Jensen and Meckling explain why decision rights should sit near relevant knowledge. Fama and Jensen explain how decision management and control can be separated. More recent AI research explains human-AI delegation, machine agency, algorithmic control and the growing autonomy of information systems.

And very recent work is beginning to recognise that agentic AI may actually redistribute organisational decision rights.

What I think we still do not understand well enough is what this looks like within large, real-world enterprises.

My hypothesis is that the issue is not simply “good AI governance” versus “bad AI governance.” Decision rights may be displaced in several distinct ways.

They may become vacant, because agentic AI creates new decisions that do not clearly belong to any existing function.

They may become decoupled, because an AI system executes a decision while accountability remains with a human who did not actually make it.

And they may migrate, because technology vendors, platforms or external partners acquire effective control over important decisions even though the formal organisation chart says otherwise.

I want to investigate whether those patterns actually exist, what other patterns emerge from the evidence, and how successful organisations reconfigure their operating models in response.

The intended contribution is therefore not another set of AI ethics principles.

I want to produce an empirically grounded theory of how decision rights and accountability change when AI systems become organisational actors.

Ideally, my research will produce three things:

First, a clearer theoretical vocabulary for describing the redistribution of decision rights.

Second, a set of enterprise operating-model archetypes showing different ways organisations govern agentic AI.

And third, a practical diagnostic framework that executives and boards can use to ask, before scaling an autonomous system:

What decisions will this system make, who has authorised those decisions, where does accountability sit, and are those arrangements actually aligned?

What makes the research particularly interesting to me is that this is not a distant theoretical problem. Enterprises are making these choices now, often without recognising that they are organisational-design choices.

The technology is moving from systems that inform the organisation to systems that can increasingly act within it.

My research question is essentially:

How must the organisation change when the system itself starts to act?

Stay tuned, if accepted, I’ll be writing a lot more about this topic!

Mr Gomez