Canada can't compete in the current paradigm - we should be looking to reshape it instead, as DeepSeek has done. Plus some quick hit reading recommendations.
My day job may bias me somewhat, but from my vantage point, DeepSeek's advance makes me much more optimistic about the potential impact of the Sovereign AI strategy as it's actually playing out.
That's _not_ to say that we shouldn't also be "actively considering how to build an inclusive and sustainable society"!
But the lessons I take from DeepSeek are that (a) small clever teams with access to modest amounts of infrastructure can still have a big impact, and (b) reinforcement learning is going to be even more important in the future than it has been.
Both of those are good for Canada (certainly better than the other model where a hand full of giants are the only ones who can advance the state of the art), and "$2 billion [...] spread across a huge range of the value chain and different stage firms" sounds a lot more promising in the former situation than the latter.
That is a fair point. I definitely think those lessons are valid. I guess my main concern is that $2 billion is being directed in a way that fits the paradigm of high-cost, high-compute, high-negative externality AI development rather than nurturing a range of nimbler, clever firms doing more efficient work. The $700 million AI Compute Challenge, for example, is trying to encourage large-scale data centre projects (with their massive environmental impacts), ensure that they are available to Canadian innovators at affordable rates, and trying to support the development of Canadian AI chips, servers, and hardware. To me, that seems to be trying to do too much with not enough funding to actually make a tangible difference, especially given if you succeed with encouraging the large data centres then you’re also going to have downstream impacts on power grids and water supply.
My day job may bias me somewhat, but from my vantage point, DeepSeek's advance makes me much more optimistic about the potential impact of the Sovereign AI strategy as it's actually playing out.
That's _not_ to say that we shouldn't also be "actively considering how to build an inclusive and sustainable society"!
But the lessons I take from DeepSeek are that (a) small clever teams with access to modest amounts of infrastructure can still have a big impact, and (b) reinforcement learning is going to be even more important in the future than it has been.
Both of those are good for Canada (certainly better than the other model where a hand full of giants are the only ones who can advance the state of the art), and "$2 billion [...] spread across a huge range of the value chain and different stage firms" sounds a lot more promising in the former situation than the latter.
That is a fair point. I definitely think those lessons are valid. I guess my main concern is that $2 billion is being directed in a way that fits the paradigm of high-cost, high-compute, high-negative externality AI development rather than nurturing a range of nimbler, clever firms doing more efficient work. The $700 million AI Compute Challenge, for example, is trying to encourage large-scale data centre projects (with their massive environmental impacts), ensure that they are available to Canadian innovators at affordable rates, and trying to support the development of Canadian AI chips, servers, and hardware. To me, that seems to be trying to do too much with not enough funding to actually make a tangible difference, especially given if you succeed with encouraging the large data centres then you’re also going to have downstream impacts on power grids and water supply.