[DEC Leadership Summit Americas] What Comes After AI Experimentation?

Higher education institutions across the Americas are not short of AI activity. Faculty are experimenting, new tools are appearing across institutional functions while pilots are producing promising results.

What many institutions still lack is a clear way of connecting that activity.

That challenge framed the DEC Leadership Summit Americas, held at Columbia University in New York on 12 September 2026. Leaders from across North and Latin America examined what it takes to move from experimentation towards a coherent institutional approach.

Experimentation Is Outpacing Institutional Strategy

Most institutions represented at the Summit remain in the earlier stages of AI adoption. In DEC’s pre-Summit survey, only 11% had reached the scaling stage.

Yet activity is already widespread. Faculty experiment with assessment, administrative teams adopt new tools, and students change how they work before guidance catches up.

As one delegate put it, “Waiting has become itself a decision.”

Decentralised experimentation cannot become the institutional strategy by default. Yet institutions cannot prescribe every use from the centre. The challenge is to provide direction while leaving space for disciplines to keep learning.

Higher Education Institutions No Longer Have a Monopoly on Knowledge

“We don’t have a monopoly of knowledge anymore in the university.”

If generative AI makes explanation, analysis and first drafts easier to access, the question shifts towards what institutions help students do with knowledge.

Can students judge its quality, apply it in unfamiliar situations and demonstrate genuine capability?

This places greater emphasis on experiential learning and stronger links with the environments graduates will enter. In parts of Latin America, these questions sit alongside wider pressures around institutional capacity, social mobility and economic development.

Faculty Change Cannot Rest on the Willing Few

AI adoption often begins with faculty already willing to experiment. Institutions are supporting these early adopters through fellowships and opportunities to share their work.

One delegate invoked a deliberately challenging phrase: “Teaching is community property.”

Teaching is not solely the property of the individual faculty member; it also sits within a department and shared curriculum. Yet much of the burden of change still falls on individual faculty.

They are being asked to rethink teaching and assessment while developing new capabilities, often alongside existing responsibilities. If institutions continue to rely on individual effort to carry that change, enthusiasm will only take them so far.

Scaling Creates a New Set of Problems

Pilots are relatively easy to contain. Scaling is not.

An initiative may work well within one course because a small team can support it closely. Extending the same approach across an institution changes the equation.

Governance becomes more complex, while resource demands that were manageable in a pilot become harder to absorb at institutional scale. New approaches to teaching or assessment may also require more faculty time. Institutions need processes that protect academic standards and manage risk, but structures built around slower cycles of change can struggle when technology develops faster than policy.

Scaling therefore becomes partly a question of what institutions are willing to reorganise. New initiatives cannot accumulate indefinitely. At some point, institutions need to decide what should change to create room for what matters.

Students Are Using AI Before Strategy Reaches Them

Students are among the first people in an institution to change their behaviour around AI, yet they are not always included in conversations shaping institutional responses.

Where institutions have created student AI councils or involved students in pilots, their contribution has gone beyond feedback. For example, students have helped refine institutional language around AI and raise issues that had received less attention from leadership.

The Summit also drew an important distinction between using AI for learning and using it for workforce productivity. Institutions need to prepare students for both, but the two require different approaches.

Institutions therefore need a clearer picture of how students are using AI and what those uses change in the learning process.

Turning Experimentation Into Institutional Change

Higher education has moved beyond asking whether AI will enter the institution. It already has.

The harder question is what institutions do with what they are learning. Which pilots should scale? What should remain locally determined? Where does governance need to move faster, and where should institutions hold the line?

These are no longer questions that can be left to individual initiatives. As experimentation spreads, institutions will increasingly need to make choices about what they want AI to change, what they want to preserve, and what they are prepared to reorganise to make that possible.

The next phase of AI adoption will not be defined by how much experimentation institutions can generate, but by whether they can turn experimentation into deliberate institutional change.