This morning Canadian entrepreneurs pursuing opportunities enabled by, and realizing, innovation are supposed to be celebrating Minister Evan Solomon's launching of Canada's "AI Strategy Task Force". I am going to present a minority position; Canada's entrepreneurs should be howling with outrage. 1: The wrong people Breakdown of task force membership by interest group represented is: Academia and advisory: 13 Entrepreneur: 7 Administration: 3 Large enterprise: 2 Capital: 1 The entrepreneurs are all growth stage companies. No one is speaking for Canada's startups. The domain where Canada's fundamental innovation issue lies, capital access, holds a single seat. A minority voice is provided to the group that matters most; customers. And those customers are large entrenched enterprises. This task force consists of the communities that have expended billions in public funds yet realized Canadian productivity ranking 28 of 38 in the OECD. This is a group whose advice should be shunned, not solicited. 2: AI is an American equities bubble, not a foundation for Canadian strategy. There is huge opportunity in application of Machine Learning (ML) to automate processes not economic to address otherwise. But this opportunity will be realized by domain specific models built and validated for domain specific applications, requiring creation of startups holding the necessary domain knowledge and engagement. "AI" at this hour is based on the thesis that broadly trained general purpose LLM's (Large Language Models) can be rapidly and affordably applied to specific commercial purposes. There was never evidence this was true, and now specific evidence that 1) "Hallucinations" (We used to call these bugs and you could get fired for them. Now they demonstrate "creativity" on the part of an LLM. Progress, I guess.) cannot be overcome, and in fact increase, with further training, and 2) the exponential compute demand created by the training being attempted to overcome the fundamental conceptual flaws inherit in LLM's, and model evaluation, mean there are few commercially viable applications for general purpose LLM's. Once again, Canada is wasting time and resources chasing last years American hype cycle as excuse to avoid addressing our own issues. 3: Cart once again placed well before horse. National wealth, and national productivity, result when innovation is applied to 1) create the competition necessary to brutally impose the need to innovate upon entrenched providers, and 2) create entirely new sources of wealth through companies addressing markets not economically serviceable by existing solutions. The specific technology is irrelevant. If AI is the best tool, great. If Magic Squirrels (tm) is the best tool, great. No tool offers strategic opportunity to Canada while the fundamental necessities necessary to realize innovation, capital access being most critical, remains unavailable in Canada.
Today, Minister Solomon announced the launch of an artificial intelligence (AI) strategy task force and public consultation to develop Canada’s next AI strategy. Building on the success of the Pan-Canadian AI Strategy, this new plan will guide Canada’s leadership in AI for the years to come. These consultations will bring together experts from industry and academia to provide actionable insights, while also inviting Canadians to share their perspectives. These discussions will focus on advancing research and talent; accelerating AI adoption across sectors; supporting commercialization and scaling Canadian AI champions; strengthening infrastructure, education and skills; and ensuring that AI systems are safe, secure and trustworthy. Learn more about how the Government of Canada is driving the responsible adoption of AI while fuelling productivity and economic growth: https://lnkd.in/gpia2ZFf #AI #AIinCanada #CdnInnovation
Key point to many LLMs. They are trained on meta data and social media, where truth is not essential and trivial topics take a large portion of the data. Using them to solve real world problems was never going to be the answer. But it is great for aspirational speeches and advertising. Core domain knowledge lies within a company's IP. Custom small language models with knowledge maps can perform much better in-house on existing computer platforms along with ML for process improvements. The notion that LLMs will give Canada its digital sovereignty back is misplaced. We do need Canadian solutions owned by Canadian companies. But it is the up-skilling of Canadians to use these bespoke custom in-house solutions is what will generate high margins and IP we can protect and augment.
They chose a news anchor who is buddies with the prime minister to develop very important policy in a field he has literally no technical knowledge expertise in.
President, Flokk Systems Inc.
3wAnd now, as a generous gift to my nation, I will save Evan Solomon six weeks and $10M and write "Canada's Next AI strategy" in 30 seconds: "blah blah blah more funding for academic research blah blah blah first nations reconciliation blah blah blah collaboration blah blah blah Canadian regulation irrelevant given ready access to US alternatives blah blah blah spend billions acquiring services from US LADS blah blah blah more study necessary blah blah blah remove inter provincial barriers blah blah blah not a nickel for Canadian entrepreneurs". Done. Now take the $10M this fiasco was going to cost and launch 10 Canadian startups. That would accomplish something to advancing Canadian innovation, productivity, and prosperity. https://www.linkedin.com/pulse/you-have-problem-mark-carney-i-solution-mark-olson-ve2dc/ Jamie McDonald: "Prioritize Proven Innovation": https://www.buildcanada.com/en/memos/prioritize-proven-innovation