In Brief
The university champions interdisciplinary answers while remaining one of the most divided institutions we have. The dominant fix, commercialisation, solves a real problem but answers a different question to the one integration poses. Generative AI now commoditises the transmissible knowledge the university has long sold, which makes the situated, accountable work it cannot own the more defensible ground. Read precisely, behaving like an ecosystem is an operating model, organised around decisions and dependencies rather than disciplines and outputs.
The previous essay closed on a question. If a single number trained policymakers to think in flows rather than systems, what have the institutions that educated them been teaching them to see? The university is the place where the habit of seeing the world in separate pieces is most carefully preserved.
Take a single river. An engineering student designs a flood scheme for it. An ecologist studies its threatened fish. An economist models the value of its water. Each works on the same system, and none needs to meet the others to graduate, publish or win a grant. The institution that most loudly insists the hardest problems are interdisciplinary is among the most divided we have built.
This is not a failure of will. It is the design working as intended. Knowledge was organised into disciplines, and funding lines, faculty structures and careers were built to reward depth within each. That was extraordinarily productive for problems that divide into parts. The trouble is that the problems now pressing on governments and communities do not divide cleanly. A catchment or a city’s exposure to heat is a coupled system where the behaviour that matters lives in the connections. An architecture built for separable problems becomes a constraint the moment the problems stop being separable.
Universities already hold what integration needs. Valuation and risk sit in business and economics. Feedback and system dynamics sit across environmental science, engineering and sociology. The work of imagining liveable futures sits in design and the humanities. What is missing is any structure that asks them to combine around a shared problem, and integration is itself a skill the institution has to build rather than a resource waiting to be unlocked.
For two decades the institutional response has been commercialisation. Innovation precincts, accelerators, spin-out funds. This solved a genuine problem, the gap between a research finding and a company. It answers a different question to the one integration poses. The commercialisation engine asks how to turn knowledge into functioning entities in the real economy. Integration asks how to turn knowledge into options a government or community can act on under constraint. For a water policy the party that has to decide is a committee under statute and budget, not a market. A fair objection is that consultancies already do this, so why a university. The answer is the combination only the university holds. It has depth across many fields under one roof and a public standing a billable firm cannot claim, held over a horizon longer than any single contract.
Synthesis is not integration
A second pressure is moving faster. Generative AI is competent at exactly the layer the university has been selling, the transmissible part you can look up, and it is felt most sharply in the classroom. The threat is real and it is clarifying, but it turns on a distinction the rush tends to blur.
Synthesis is pulling together what many disciplines have said into a fluent account. AI is good at this and getting better, and that is worth conceding plainly. Integration is different in kind. It means holding frames that do not reduce to a common measure, a priced risk and a cultural obligation that cannot sit on the same axis, and moving deliberately between opening a problem up and closing it to a decision. A system built to predict the most probable continuation has a structural tendency to do the reverse. It settles on the average framing and smooths away the very incommensurability that integration exists to hold. The appearance of synthesis is not the act of integration.
Two further limits are structural rather than temporary. A predictive system cannot be the party answerable for the choice, and it cannot hold the trust and legitimacy that let a committee act on a contested call. A system can read everything ever written about the river. It cannot sit in the room where the decision is made, weigh the competing interests and own the outcome. That situated, accountable work is where a university’s value is moving, and it is exactly what an ecosystem-shaped institution would put at its centre.
What an ecosystem organises
The core idea here is not new, nor is it flimsy. Ronald Barnett argued more than a decade ago that the university should take its own interconnectedness seriously and act with an ethic of care. Used precisely, the metaphor points at something specific. An ecosystem is not a catalogue of species. It is a structure of flows and dependencies, and some elements matter out of all proportion to their size. Ecologists call these keystones. The translation for a university is that an ecosystem organises around relationships and the decisions they force, not around an inventory of disciplines.
Three moves follow. Organise around the decisions partners face rather than the disciplines researchers belong to. Value inquiry on its ambition as well as its readiness, so genuinely new questions are not screened out for being early, which is also the safeguard against being pulled toward near-term client work. Ask, every time, who is affected by a decision but absent from the room.
These core elements are what make the dependencies visible. To return to the river, a university running a dozen separate projects scores each on its own merits. Mapped against one another, a shared account of how that river system behaves turns out to underpin most of the others. It is not the eye-catching project, but it is the keystone. Build it first and the rest stand on something solid. Naming the keystone is still a judgement that can be made wrongly, but the dependency map makes the reasoning explicit enough to be challenged and corrected.
A clean example of this approach is the Western Australian Biodiversity Science Institute (WABSI), which starts from the knowledge gaps decision-makers actually face. The honest caveat is that WABSI is a purpose-built joint venture, not a university. It shows what the principle looks like but not that a large institution, weighed down by its own mass, can reach it. That is the unsolved part. Several Australian universities are building cross-cutting institutes, RMIT’s Regenerative Futures Institute and ANU’s climate institute among them, but these integrate around curriculum and capability rather than around decisions. The decision-context model is a design principle with a test attached, and the way to begin testing it is not to reorganise a university but to run it once, inside a single centre.
The fragmented university sees a dozen separate studies of the river. The commercialised university sees a spin-out. The university that behaves like an ecosystem sees a decision someone has to make about a living system, and organises what it knows around getting it right. That is the value an automated tutor cannot reach, which is why the shift is no longer optional.
There is a further wall, and it is the most important one. Breaking down the barriers between economics and ecology still leaves the institution inside a single tradition of knowledge. To behave like an ecosystem in any serious sense, the university has to learn to work across knowledge systems, not only across its faculties. That is where we turn next, in “Stewardship as Method.”
Read the full essay, with practical exercises, references and further reading on emerdigm.com → https://www.emerdigm.com/essays/the-university-as-an-ecosystem