Groundbreaking ideas and research for engaged leaders
Rotman Insights Hub | University of Toronto - Rotman School of Management Groundbreaking ideas and research for engaged leaders
Rotman Insights Hub | University of Toronto - Rotman School of Management

Leadership forum: investing in disruption

Read time:

Ajay Agrawal, Jüergen Schmidhuber, Shivon Zilis, Patrick Wood, Alex Shevchenko, Charles Plant

Ajay Agrawal

Founder, Creative Destruction Lab and Machine Learning and the Market for Intelligence Conference; Geoffrey Taber chair in entrepreneurship and innovation, Rotman School of Management

“You can see the computer age everywhere but in the productivity statistics.” So stated Nobel Laureate and MIT economics professor Robert Solow in 1987. Eventually, economists found where the productivity gains from the computer age were hiding: in the future. While they eventually showed up, they took longer than expected because they were tied to investments in complements — all of the things other than algorithms/models that are necessary to make commercial-grade AI work (data, redesigned workflows, training, regulation, human judgment, infrastructure, etc.).

As in the computer age, the widespread productivity gains associated with machine intelligence will depend on investments in complements. As we shift from technical achievements in AI (‘Look everyone! The AI can read a handwritten address on an envelope! The AI can drive a car! The AI can classify a medical image!’) to large-scale commercial deployment, the design and implementation of complements will be paramount.

The computer scientists designing AIs are far ahead of those building the complements — industry practitioners, social scientists, regulators and the like. Now that everyone has realized the sweeping potential of AI, companies and countries are racing to create and control the complements. While the algorithms are software and thus have low barriers to entry (notwithstanding scale advantages with respect to training data), many complements require significant capital expenditure and thus have higher entry barriers. Therefore, competition policy and market dynamics will move even further onto centre stage.

In other words, we are entering the next phase of the AI revolution: competition in the market for AI complements. This will feel different from what we’ve experienced so far. The genteel competition among computer scientists on display at conferences like NIPS that is based on the performance of new AI algorithms against well-specified technical benchmarks like ImageNet will give way to competition among firms over the ownership and control of scarce complements such as data, infrastructure, talent and relationships.

For enterprises, competition in the semi-scientific culture of algorithmic performance against benchmarks was curious and novel. However, competition over complements is familiar territory.

And given the size of the prize, this competition is likely to get rough and tumble, as corporate AI strategies depend at least as much on complements as algorithms. Intensified competition will increase the pressure on companies to deliver results. Internal debates like the one at Google regarding whether to abandon Project Maven — a collaboration with the U.S. Department of Defence to utilize AI for image analysis that could potentially be used to improve drone strikes — will seem quaint. Furthermore, competition will not only intensify at the company level. In recent months, one country after another has announced its national AI Strategy — and most of them read more like industrial than science policy. Competition over complements is about to become fierce.

Jüergen Schmidhuber

Co-founder and chief scientist, NNAISENSE; scientific director, Dalle Molle Institute for AI Research; professor of AI, University of Applied Sciences and Arts of Southern Switzerland

Even minor extensions of existing machine learning and neural network algorithms will achieve important super-human feats in numerous fields, ranging from medical diagnostics to smarter smartphones that will solve more of your problems. In the not-too-distant future we will have what I call watch-and-learn robotics, where we quickly teach an artificial neural network to control a complex robot with many degrees of freedom to execute complex tasks, such as assembling a smartphone, solely by visual demonstration and by talking to it, without touching or otherwise guiding the robot. Going forward, this will affect many professions.

My company recently won the Learning to Run competition at NIPS, the machine learning conference in Long Beach, California, going up against over 400 competitors from industry and academia. The challenge was to learn to control the muscles of a simulated human torso and make it run as far as possible without a teacher. Human babies need a year or so to learn to get up and walk and run, and our runner also needed many weeks of computation time.

Watch this video: Teaching Artificial Intelligence to Run (NIPS 2017)

Elsewhere, in a project with Audi, we built the brains of the first model cars that learned how to park from scratch, again without a teacher. We also have AI applications in industry and finance. Our joint venture with Acatis is called Quantenstein, with the first purely AI-driven portfolio management.

Watch this video: Learning to Park | A collaboration between NNAISENSE and Audi Electronics Venture

We believe we can go far beyond what is possible today and pull off the big practical breakthrough that will change everything, in line with my motto since the 1970s: Build an AI smarter than myself, so that I can retire. What would it be used for? Everything. Humans should do zero per cent of the hard and boring work, computers the rest. How far off is this? Not more than a few years or decades. We are currently witnessing the ignition phase of this field’s explosion. This is much more than just another industrial revolution. This is something new that will transcend humankind and even biology.

People often ask me how to prepare for the future. I tell them what I tell my daughters: Be prepared to learn new things all the time, and make sure to learn how to learn. It will always be an advantage to know a bit of math and physics, because the world is based on them. In addition to that, acquire something that I don’t have — namely, social skills. Then you will be able to quickly master all new challenges.

Shivon Zilis

Project director, office of the CEO, Neuralink & Tesla; co-chair, Machine Learning and the Market for Intelligence Conference @ Rotman

Organizations like Google, Facebook, Apple, Microsoft, Amazon, Uber and Bloomberg bet heavily on machine intelligence, and its capabilities are pervasive throughout their suite of products. But for most organizations, the successful use of machine intelligence is surprisingly binary — like flipping a stubborn light switch. Why is it so difficult for companies to wrap their heads around it? Because machine intelligence is different from traditional software. Unlike with big data — where you could buy a new capability — machine intelligence depends on deeper organizational and process changes. Companies need to decide whether they will trust machine intelligence analysis for one-off decisions or if they will embed machine intelligence models in their core processes; teams need to figure out how to test newfound capabilities; and employees need to be coached to learn from the data they enter.

Unlike traditional hard-coded software, machine intelligence gives only probabilistic outputs. We want to be able to ask machine intelligence to make subjective decisions based on imperfect information. As a result, machine intelligence software will make mistakes, just like people do, and we’ll need to be thoughtful about when to trust it and when not to. The concept of this new machine trust is daunting and makes machine intelligence harder to adopt than traditional software. I’ve had a few people tell me that the biggest predictor of whether a company will successfully adopt machine intelligence is whether they have a C-Suite executive with an advanced math degree. These executives understand that this isn’t magic — it is just (really hard) math.

I see all of this activity only continuing to accelerate. The world will give us more open sourced and commercially available machine intelligence building blocks; there will be more data; there will be more people interested in learning these methods — and there will always be plenty of problems worth solving.

Patrick Wood (UofT BA 94)

President, Tormont Group and founder, The Tormont50

These days, the big investment banks are generally looking for easy kill. They tend to gravitate towards companies that are easy to fund over those that would take more time and effort.

We focus on finding the smaller opportunities that they overlook. The Tormont50 platform features growth opportunities in the small and microcap space that we believe have significant growth potential for institutional investors, and we open it up to a base of about 75 institutions across Canada and the U.S. To be included, a small-cap company has to have identifiable growth drivers, whether it be an impressive innovation or the potential disruption of an industry. Even a change in management can have a broad impact on a company’s growth. We look out for everything that can make a company a future winner.

In Canada, the companies getting the most funding right now include cannabis companies, to the detriment of other segments. Rather than focusing on the growers, we are looking at all the verticals that go around that. In my view, investors should be a bit skeptical about the rapid rise of Canadian cannabis companies, because meeting expectations will likely be a big challenge for many of them. There’s a lot of chatter on Bay Street and Wall Street about that right now. The U.S. is a much more mature market: Colorado and Washington have been legalized for several years now, and as a result, software, branding, distribution — all the key things that are missing in the Canadian market space — have really been perfected there.

We’re really excited about Blockchain right now, and we’ve brought companies like Delphx to Canada through the listing process. It’s one of the pure-use cases of Blockchain and smart contract technology. Not many of these companies have been represented in Canada yet, so we’re actively seeking them, particularly in the U.S., which is generally a much more mature market by virtue of economies of scale. In terms of other high-growth industries, crypto had a good run, but it was brief and based purely on speculation, and it officially ended recently. Software will continue to be a clear winner, and we’re happy to be presenting smaller, lesser known companies to the Canadian markets.

With respect to Blockchain and AI companies, Canadian opportunities carry with them a lot more risk than their U.S. counterparts. We’re really far behind what is happening in the U.S. AI is a part of every company we work with, but we have yet to come across a strong enough pure-play AI start-up that would make us excited about taking it through the process of listing, introductions to investment banks, institutional investors, etc. Some of the larger companies like IBM and Microsoft have really taken over that space because they were there from the start. From a smaller company perspective, we’ve considered a few opportunities, but nothing that would justify us wanting to go deeper. We are certainly looking out for these companies. Hopefully, they will emerge soon.

Alex Shevchenko

(Rotman MBA 06), founder and product manager, Grammarly

Both my parents were university professors, and I became fascinated by educational technology early on. My first company, MyDropbox.com, was designed to help educational institutions tackle the problem of plagiarism. My most important takeaway from that experience was the critical importance of focusing on solving a problem that is real, that many people have and that you really understand. After selling MyDropbox.com [to Blackboard Inc.], Max Lytvyn, Dmytro Lider and I set out to solve another real-world problem: to help people communicate better in English.

As an ESL speaker myself, I had witnessed firsthand how people like me miss out on opportunities due to a struggle to write correctly in English. Our vision for Grammarly was to apply artificial intelligence to help billions of people around the world communicate more accurately and clearly. We started with grammar, spelling and punctuation, and today, Grammarly also addresses the clarity and effectiveness of our users writing. We provide feedback on issues such as wordiness, vagueness and hedging; word choice (including suggestions for more inclusive language); sentence structure and — yes — plagiarism.

We recognized early on that the market for this problem is enormous: There are close to two billion people in the world communicating in English, and the vast majority of them can benefit from our product. Our more than 10 million daily active users represent a wide variety of people: They are students, professionals, job seekers and English language learners. They include all sorts of professionals, from engineers to marketers to journalists who write for a living. Everyone can use help from time to time.

Grammarly was actually bootstrapped for the first seven years, and I’m proud to say that we were already profitable when we raised our first financing in 2017. Very early on, we made a conscious decision to delay seeking external funding for as long as possible. We believe that the best possible source of funding is always your customers, so we focused on building the best possible product to delight our users.

We were extremely selective when choosing investment partners. It was important for us to work with people who shared our mission and values. The funding process took some time, but it was well worth it. We ended up in complete alignment with our investors, without having to compromise who we are. We’re honoured to have General Catalyst, Breyer Capital, IVP, Signal fire and Spark Capital — some of the most prominent VC firms in Silicon Valley — among our investors.

Putting together an effective founding team is really difficult because it’s so important that the initial people have complementary skills and work together effectively. Making early hires is one of the biggest bets a start-up will ever make, and the process can’t be rushed. To this day, we maintain an exceptionally high hiring bar.

Another important and difficult thing is staying focused, which is much harder in practice than it sounds. It is really important to keep your eyes on the ball with a clear strategy. My advice is to do fewer things, but do them extremely well so you can focus on execution.

Charles Plant

Serial entrepreneur and senior fellow, Impact Centre, University of Toronto

In Canada, our start-ups rank very high in the world on number of exits, but we don’t rank very high on the value achieved from those exits. When you look at the return rate for our venture capitalists, it’s pretty low in comparison to the U.S. — although it has improved in the last few years, particularly since 2008. We are now having more successful exits by selling them later, after higher growth and bigger revenue, so we’re getting more money for them; but we can do better.

People frequently complain that there’s not enough money for entrepreneurs in Canada, but I don’t believe that. The fact is, high-growth companies will attract capital from anywhere in the world. If our companies aren’t getting that capital, it indicates that there aren’t enough of them growing fast enough to attract large-scale capital. Our biggest challenge is to figure out how to grow companies a lot faster.

In my view, we’re not in large-enough markets — not just geographically but perceptually. For instance, we don’t tend to start consumer-based companies. Many large-scale tech companies are consumer-based: Uber, Airbnb, Facebook, Amazon, Google, Apple. We have BlackBerry, but we just don’t start enough companies in consumer-based markets. The minute you ignore those markets, you cut your growth potential down significantly.

So, what has to change? First, we have to understand what it takes to scale. We should be trying to grow companies by a minimum of 100 per cent a year. Second, we’ve got to apply capital to that, which means we have to change the way our institutions capitalize. We have too many small funds and not enough large ones that can write big cheques. Third, we lack the required marketing and sales talent here, so our markets should be foreign and we should be hiring foreign marketing and sales expertise in those markets — people who understand the nuances of a market. The combination of these three things, I think, would have a great impact on the growth of Canadian start-ups.

Editor's note: The Impact Centre’s reports, “The Class of 2008”, “A Delicate Balance” and “The Land of Stranded Pliots” can be downloaded at www.impactcentre.ca/discover


This article originally appeared in the Winter 2018 issue of Rotman Management magazine.

Ajay Agrawal, founder, Creative Destruction Lab and Machine Learning and the Market for Intelligence Conference; professor of strategy and Geoffrey Taber chair in entrepreneurship and innovation, Rotman School of Management
 
Jüergen Schmidhuber, co-founder and chief scientist, NNAISENSE; scientific director, Dalle Molle Institute for AI Research; professor of AI, University of Applied Sciences and Arts of Southern Switzerland
 
Shivon Zilis, project director, office of the CEO, Neuralink & Tesla; co-chair, Machine Learning and the Market for Intelligence Conference @ Rotman
 
 Patrick Wood (UofT BA 94), president, Tormont Group and founder, The Tormont50
 Alex Shevchenko (Rotman MBA 06), founder and product manager, Grammarly
 
Charles Plant, serial entrepreneur and senior fellow, Impact Centre, University of Toronto