Proposal in development
This page describes a proposal. There is no West Coast Centre of AI Excellence. No government has designated one, no program funds one, and no organization has agreed to join one.
BC + AI is developing the case for what a West Coast model could do, and publishing it while it is still early enough to argue with. A proposal that only circulates privately gets agreed to by the people already in the room. This one is written down so the people who would have to build it can tell us where it is wrong.
BC + AI is in active conversation with federal and provincial officials about what a West Coast model could look like.
Why British Columbia, why now
Canada’s national AI strategy, AI for All, includes a $500 million investment to expand the Regional Artificial Intelligence Initiative, which is delivered through the regional development agencies rather than through a single national body. The Canadian Sovereign AI Compute Strategy runs alongside it. Both are federal instruments that regions are expected to have a position on.
Having a position means more than wanting the money. It means somebody in the province can say what the capability gap is, who would close it, what would be built, who would be accountable for it, and what happens to the people the technology lands on. Those are not questions any one organization in the province has been resourced to answer end to end, and this proposal starts from the assumption that somebody should be.
The argument for doing this now is timing, not urgency theatre. Programs of this size set their shape early. The regions that show up with a worked-out model tend to get a model closer to the one they wanted.
The gap
Innovation, Science and Economic Development Canada describes Amii, Mila and the Vector Institute as Canada’s three national AI institutes. They were built under the Pan-Canadian AI Strategy to concentrate research talent, and by that measure they worked.
British Columbia has no institute in that group. That is a statement about where three organizations are located, not a grievance. No federal process to designate another national AI institute has been announced, and this proposal is not an application to one.
The live federal instrument aimed at regions like this one is the Regional Artificial Intelligence Initiative. So the useful question is not how British Columbia gets added to a list. It is what kind of body the province actually needs, and whether that body looks anything like the three that already exist.
A different West Coast model
A West Coast centre would not be a copy of a research institute placed on the coast. British Columbia’s AI work is spread across film and games studios, health authorities, ports and resource operations, municipalities, school districts, First Nations, small companies and a lot of individual practitioners. A single building in a single city would miss most of it.
The proposal is for something distributed: a coordinating body with real capacity in more than one region, connecting universities, industry, public institutions, Indigenous partners, community organizations and the networks that already exist, rather than replacing any of them.
That shape has an obvious failure mode. Distributed can mean diffuse, a logo on other people’s work. Avoiding it is a design problem the proposal has to solve before it is worth funding, and it is one of the things this page is published early to get argued with about.
Indigenous participation and governance should be designed with Indigenous partners.
What a centre could do
BC + AI runs community-scale versions of several of these functions today. That is not the same as operating any of them at provincial scale, and this proposal should not be read as a claim that it does. Community scale is how you learn what the work requires. It is not the work.
01 / Responsible AI adoption
Responsible is a word that means nothing until someone names the responsibility. Here it means a named person is accountable when a deployed system harms someone, there is a route to challenge a decision, and somebody has costed the fix before the launch date. Most adoption support in the market helps organizations buy faster. This function would also help them stop.
02 / Workforce and public AI literacy
The workforce question in British Columbia is less about producing more researchers than about the nurse, the city planner, the animator and the mill supervisor who now have these tools in their job. Public literacy is the same problem one step out: people are being governed by systems they were never shown. Both need delivery over years, in more than one part of the province.
03 / Applied research, development and experimentation
Applied means the problem comes from a health authority, a studio, a Nation, a port or a small company, and the output is something they can use. It also means the failures get written down. A centre that only publishes its successes is a marketing function with a research budget.
04 / Regional and cross-sector coordination
This is the least glamorous function and probably the highest return. Right now a school district, a health authority and a municipality can each spend a year on the same procurement question without ever meeting. Coordination is not a conference. It is somebody whose job is to know who is already doing the thing you are about to start.
05 / Public-interest governance
Consultation usually arrives late, in English, as a survey, with six weeks to respond. A body that convenes properly can produce something better: a documented public record of what people in the province actually said, available to anyone who wants to check whether the final decision reflected it. Governance of the centre itself is part of this function, not separate from it.
06 / Sovereign and sustainable AI infrastructure
This is an access and governance function. Allocation policy, public-interest terms, environmental and energy accounting, and the question of what a publicly supported cluster is permitted to run. Nothing in this proposal has BC + AI owning or operating a data centre. Owning hardware carries capital, regulatory, security and operating obligations that have nothing in common with running an association, and the leverage worth having here is over the terms, not the racks.
Built to complement, not duplicate
Innovation, Science and Economic Development Canada names Amii, Mila and the Vector Institute as Canada’s three national AI institutes. They were funded under the Pan-Canadian AI Strategy to attract, train and retain research talent, and they built deep research concentrations in Edmonton, Montreal and Toronto.
Nothing in this proposal is a claim that those institutes failed British Columbia. They were built to do research and train researchers. Running adoption programs, workforce delivery and public governance work in every province was never their mandate, and a national network reaches a coast when somebody on the coast builds the thing that connects to it.
The national institutes concentrate research and graduate training. The functions on this page are mostly adoption, literacy, coordination and governance. A West Coast centre would be a customer and a collaborator of that research, not a rival bidder for it.
Distance from Toronto, Montreal and Edmonton is the point rather than the complaint. A body based here would work with the sectors that concentrate on this coast, and on Pacific rather than eastern time and trade orientation.
The cheapest failure available is to rebuild three institutes badly somewhere else. Anything a national institute already does well should be routed to it. A West Coast centre earns its existence on what is missing, not on what it can duplicate.
What already exists
When Canada consulted on AI transparency, a BC + AI room worked the federal questions together and published the response, the recording, the transcript and a disclosure of how the document was written. That is the governance function at community scale, with the record left behind so anyone can audit it.
The AI transparency responseBC + AI has published what it thinks Canada’s AI strategy should account for. It is published under the association’s name, with a date on it, which is the minimum standard for anyone asking to be taken seriously in this conversation.
Our platform for the AI task forceThe public compute project is where the access and governance argument in function 06 comes from. It is a live, in-progress record rather than a finished position, and it is published in that state deliberately.
The public compute projectRegular gatherings in three regions of the province, running since 2023, with the archive public. This is the coordination function at the scale an association can currently sustain.
The communitiesNone of this is a centre. It is the evidence that convening capacity exists in the province at community scale, and the reason BC + AI thinks it is worth arguing about what a larger version would look like.
Who needs to be at the table
No organization has agreed to any of this, and none is named here. Publishing a list of names before those conversations happen would make the proposal look further along than it is.
Participation and governance designed with Indigenous partners, on terms set in those conversations rather than in this document.
Research capacity, applied programs, and the training pipeline the workforce function depends on.
Mandate, public service delivery, and the policy surfaces where adoption decisions are actually made.
The instruments that already exist, and the question of whether a West Coast model fits any of them.
The public institutions deploying these systems now, often making those decisions without dedicated support.
Two different constituencies with two different problems. Both have to be at the table and they should not be counted as one.
The people whose work changes first, represented as participants rather than as a consultation line item.
Public literacy delivery, and the constituencies that carry the cost when adoption goes badly.
The people already doing this work locally, including the ones who would be skeptical of a centre and should be.
Research and background
Innovation, Science and Economic Development Canada
The current federal strategy, and the frame every argument on this page has to survive contact with.
Innovation, Science and Economic Development Canada
The strategy that built the national institutes, and the page that names Amii, Mila and the Vector Institute as the three of them.
Innovation, Science and Economic Development Canada
The federal compute program, and the reason function 06 is written as an access and governance question.
Innovation, Science and Economic Development Canada
Delivered through the regional development agencies, including PacifiCan. The live federal instrument pointed at regions like this one.
Get involved
Public and institutional partners
If you work in government, a public institution, a Nation, a post-secondary or a health authority, the most useful thing you can do with this page is disagree with it specifically. What would a body like this have to do before it was worth your time, and what would make you want nothing to do with it?
Industry and innovation partners
If you build or adopt AI in British Columbia, the functions on this page are a guess about what you are missing. Which ones would change how your organization works, which are already covered by somebody else, and what did we leave out entirely?