Jake Hirsch-Allen's Recent LinkedIn Posts

The Public AI drumbeat continues Ethan Phillips argued yesterday that Canada needs a publicly-owned generative AI model at the centre of Minister Evan Solomon’s upcoming strategy. He echoes arguments from the our team at The Dais at Toronto Metropolitan University and allies like Nicholas Vincent, Mark Surman, Joshua Tan, Nicholas P. Garcia, Luca Cominassi, Beth Simone Noveck, David Eaves, and many others, that Switzerland’s Apertus is the proof of concept Canada should build a middle power coalition around and turn into a model for the world. Public AI is AI as public infrastructure — provisioned like electricity, water, roads, libraries, or the internet itself. Publicly accessible. Canada’s $2.4B in sovereign compute is promising but partial. Without open source principles baked in from the start — interoperable, standards-based, built so that improvements in one country or lab accrue into a shared pool — we are building infrastructure that will eventually be captured or locked down. Publicly accountable. A public model is only as trustworthy as its data. Ana Serrano is right that sovereignty is not just about servers on Canadian land — it is about the culture of how AI is built and shared. We need a Canadian data and AI commons with benefit-sharing protocols, participatory governance, and consent-based data, bilingualism, Indigenous data sovereignty, and Charter values. Permanent public goods. Canada has lead the world when the stakes were high such as with public health care. The CBC/Radio-Canada of AI is not a fantasy. It is a policy choice other are making and we should be leading. Switzerland launched Apertus. Germany is building SOOFI. Canada should lead the coalition of values-aligned democracies as advocated by Jon Shell and Mark Carney among many others and one focus should be Public AI. We have done this before. Lloyd Axworthy, PC, CC, OM, Ph.D. did not wait for Washington to ban landmines, establish the International Criminal Court, or articulate the Responsibility to Protect. He built a middle power coalition, moved at the speed of values, and changed international law. Public AI should be built by this generation’s equivalent coalition. An alliance of governments, research institutes, open source communities, and civil society organizations — none of whom can match big tech or China alone — can pool compute, data, talent, and governance standards. Canada, through organizations like CIFAR has led this kind of coalition before. Canada’s diversity, its history of multiculturalism, its Indigenous peoples, and its good governance are differentiators. No other country in the G7 is as well positioned to lead the alliance that builds AI that is consent-based, multilingual and pluralistic. What is missing is the political will to make it accessible to everyone and accountable to the public. Adoption moves at the speed of trust. Trust requires democratic accountability over the systems reshaping our lives.