AI Geekly - Expert Opinions

Hearing from AI Leaders

Expert Opinions

Welcome back to the AI Geekly, by Brodie Woods. This week we’re switching things up (he says that every week), featuring an interview with Simon Dandavino, MBA, PhD, Director of Programs and Partnerships at Canada’s Next AI, a leading national accelerator and founder development program for early-stage AI-powered startups. Next AI focuses on identifying talented idea or early-stage teams with ambitious solutions and provides mentorship, education, and a network to disrupt industries.

TL;DR Convos with Leaders in AI; Figure 01 Robot gets ‘Wicked Smaht’; Cohere Loud and Clear; GPU Poors’ Rags to Riches

For those not in the know, Canada is one of the most vibrant (though capital constrained) AI hubs globally, note for example that three of the four so-called “Godfathers of AI” are either Canadian or faculty members of Canadian universities. As we’ll cover in our upcoming AI 101 Primer piece, the foundational work by Geoffrey Hinton et al. resuscitating and reinventing neural networks with modern technology is largely responsible for the advent of today’s AI models. We’ll close out with a quick peek at some of the most interesting developments this week, including new robots with AI brains, a new model from Cohere, and some impressive developments in AI training and inference that make consumer GPUs viable —a boon for the Open Source movement and the democratization of AI.

Standing on the Shoulders of Giants
Simon Dandavino @ Next AI on Education and Learning

Brodie - AI Geekly (Q): How do you see the role of formal education evolving in the AI field, especially with the rapid pace of technological advancement?

Simon - Next AI (A): I think that the impact of AI will be felt much beyond what would be considered the field of AI. Already, formal education was shifting from a mission of teaching knowledge to a mission of teaching aptitudes. The world is changing fast, and the knowledge that is relevant when you are studying will likely not be the one you need as part of your career or, more accurately, careers. In that sense, the role of AI, and I’m referring particularly to LLMs, will be to act as gateway to knowledge, possibly with an impact equivalent to the arrival of the internet. Regarding education of AI itself, it is difficult to tell. There is no doubt that there will continue to be advanced research and education in the field, but a lot of people will become advanced AI users without really understanding the core. In the same way that you don’t need to know machine language to code, you won’t need to know the model development (emphasis added). We see a lot of startups using open libraries that have not developed their own models but are building their differentiation on the other parts of their system architecture.

Brodie - AI Geekly (Q): With a background in both engineering and business, how important do you think interdisciplinary knowledge is for emerging AI professionals and entrepreneurs?

Simon - Next AI (A): Hugely important. The strongest teams are the ones where there is alignment on mission and values, but diversity in skillsets and approaches. Too often, we see amazing tech entrepreneurs with great skills, ideas and terrible business sense, or amazing entrepreneurs who lack the tech skills to implement their vision. We try to get these communities to talk, but it’s not easy! For me, probably the most important ingredient to the creation of new ventures is multidisciplinary opportunities in the ecosystem where founders can discover each other.

AI Meets World
Simon Dandavino @ Next AI on Startups, Ethics, and the Future

Brodie - AI Geekly (Q): As Director of Programs & Partnerships at Next AI - Montréal, what are the key factors you consider when selecting AI startups for the program, and how do these factors align with current trends in AI commercialization?

Simon - Next AI (A): We have always had a program that was focused on commercialization. As an early-stage accelerator program, we typically evaluate ventures on three dimensions: the technology, the team and the market potential. On the team aspect, we include subject matter expertise, which is a key success factor. In terms of trends, we of course see a lot of Gen AI this year. It’s interesting how this is bringing back some sectors, like education, that were a bit less present in the last years. I think that service industries, which are human interaction driven, will see a lot of use cases that use generative AI. Going back to the last question, we also see more diversity in ideas, since the barrier to entry has lowered and more people want to find ways to solve the problems that they see around them.

Brodie - AI Geekly (Q): In your view, what are the most pressing ethical and governance challenges in AI today, and how is Next AI preparing its participants to address these issues?

Simon - Next AI (A): I’ve been concerned about deepfakes for several years now, and the recent advances have only made things more scary in that sense. Impacts can be wide ranging, from discrimination to cyberintimidation. I’m particularly worried about teenagers who also have enough on their plate when it comes to social network and peer pressure. Beyond this, there will be major disruption on the markets. A lot of jobs will be replaced by AI, and I’m not convinced that just as many will be created. And even so, the created jobs might be less accessible to a lot of people than the old ones. Our society will have to shift to adapt to this. My hope is that with proper social support, we could go towards a society where people work less to do the same quantity, and keep more time to take care of themselves and be part of their communities. As part of our program, we always include some education on the ethical use of AI.

Brodie - AI Geekly (Q): As we look ahead through 2024, what emerging trends or technologies in AI do you find most exciting, and why?

Simon - Next AI (A): I think we’re just seeing the beginning of generative AI, and video generation is rolling out at a rapid pace. An optimistic view of Gen AI is that it is a creativity empowerment tool. Many people, who do not have the skills to be content creators can now be, being only limited by their imagination. I find this fascinating as I believe creativity and imagination are such basic human characteristics that we put aside as we become adults. When I was on midjourney in the early days and was seeing all the crazy images that people were having generated, I had a sense that there was finally a release… that people could put in images and share what was until then trapped in their mind.

The Holdovers
Figure 01 bot gets smarter, Cohere conceives, and Democratization of AI gets a serious boost

Figure 01, the robotics company that previously announced a collab with OpenAI released the demo above of its robot imbued with OpenAI’s vision language AI model de a previously unseen level of awareness and understanding in the highly scripted video. This is in keeping with our 2024 thesis re: convergence of AI advances with those in robotics. Expect more from these guys.

Cohere’s new highly scalable enterprise-focused Command-R model deserves a shout-out. In our experience, few vendors, particularly in the burgeoning AI space, “get it” when it comes to enterprise-level requirements. Cohere’s Command-R offering shows that they don’t just “get it”, they “have it”. Designed with Retrieval Augmented Generation (RAG) and tool use in mind the model can retrieve necessary documents to ground its knowledgebase and perform actions. Add-in 128k context window, low latency and price, and open model weights (yay transparency) and this is a solid contender for enterprise use cases.

Answer.AI and Together.AI both did a great service to the Open Source community and its overarching goal of democratizing AI. Answer has released a detailed blog post on the process it developed to train new 70B parameter AI models (hint: that’s big) on relatively affordable consumer hardware (graphics cards/GPUs), something that was essentially impossible up to this point —so a huge boost to the training half of the AI equation for open source users. Inference, the other side of the AI coin, got a boost of its own via Together’s new Sequoia model which dramatically accelerates inference on consumer GPUs. We won’t go into the gory details about model quantization or speculative decoding, but trust us when we say this is cool stuff.

Before you go… We have one quick question for you:

If this week's AI Geekly were a stock, would you:

Login or Subscribe to participate in polls.

About the Author: Brodie Woods

With over 18 years of capital markets experience as a publishing equities analyst, an investment banker, a CTO, and an AI Strategist leading North American banks and boutiques, I bring a unique perspective to the AI Geekly. This viewpoint is informed by participation in two decades of capital market cycles from the front lines; publication of in-depth research for institutional audiences based on proprietary financial models; execution of hundreds of M&A and financing transactions; leadership roles in planning, implementing, and maintaining of the tech stack for a broker dealer; and, most recently, heading the AI strategy for the Capital Markets division of the eighth-largest commercial bank in North America.