Industry Insider Series: Patrick Surry, Christopher Berry, Josh Lessing
Welcome to our Industry Insider series, where we ask our faculty of expert speakers to weigh in on industry trends, innovation and more. Make sure to subscribe to our newsletter for even more Q&A’s with our Big Data and AI Toronto speakers.
Patrick Surry leads data science at Hopper. He is a leading global analytics practitioner with a wealth of experience in real-world delivery of customer insight, predictive analytics and behavior modeling.
Josh Lessing is one of the world’s leading minds on dexterous robotic handling. His current venture, Root AI, is integrating advanced robotics, vision systems and artificial intelligence to automate the multibillion dollar agricultural sector.
Christopher Berry leads product intelligence at the Canadian Broadcasting Corporation (CBC) where he and his teams turn data into product.
How did you start a career in Big Data / Artificial Intelligence?
JL: I started out my professional career in robotics, specifically robotics for food. As I learned more about the challenges that we’re all going to face in producing food for the future, I realized in order to create more food, you need to manage plants on an individual basis. While this is fundamentally impossible for a human to do, it is immensely possible and even inevitable for artificial intelligence. At Root AI, we’re working on leveraging plant level data & AI to create actionable intelligence for growers.
PS: My career in Big Data began before we started calling it Big Data! At the core of it, I’ve always been interested in problem-solving and how things work. A love of math and some self-taught computing exploration got me into applied math (simulating complex systems like weather, airflow, particle physics, etc.) which led to parallel computing and optimization algorithms, which in turn led to evolutionary algorithms and working on some early neural network models. In trying to apply these to real-world problems, like those of banks and retailers, I realized that visibility, accessibility and the accuracy of data is more important than any one single algorithm, which got me into consulting and working on building data analysis products. Eventually, I found Hopper, where the data science team and I have the amazing opportunity of having data as our product. Our job is to help consumers make smarter decisions based on really complex data and statistics – which means our challenge is to establish trust with our users without overwhelming them with complexity and then reduce the overall stress of their travel purchase experience.
CB: I discovered a passion for technology, applied statistics, and public policy in university. It’s never been about tech for the sake of tech, or stats for the sake of stats, or big public policy for the sake of big public policy. These are tools to achieve some outcome. I played a lot with genetic algorithms in university to achieve outcomes, so when data science emerged as a real movement in 2009, and a lot of mathematical statistics got rebranded as artificial intelligence around the same time, I was relieved to find that I wasn’t alone. And I was ready for it.
Those outside the tech industry have a difficult time defining exactly what AI is. We’d like to know: What isn’t AI?
PS: AI, like the term “big data”, has become a buzzword that means a lot of different things to different people. Originally, AI was all about making machines that could think like humans, and although some methods (like neural networks) are loosely modelled on neurological processes, most of what is now described as “AI” isn’t about that.
To me, AI is more about describing a broad set of computer algorithms that are good at learning to generalize patterns for specific prediction problems, like whether a flight price is likely to go up or down based on the previous prices we’ve seen for that, and related, trips. Machine Learning is a much more descriptive term for what we’re actually doing.
JL: AI is not a replacement for people… it’s an enhancement for human capabilities. Modern AI sifts through mountains of data and elevates the most salient, actionable intel to the user. For instance, growers have a profound level of understanding of crop health but do not have the resources to care for each plant on an individual basis. Big data + artificial intelligence + robotics will create a system that leverages the complete experience of a master grower and utilizes that knowledge to care for plants on an individual basis.
CB: The statistical techniques applied to a set of very narrow problems is not general AI. The bulk of the industry is focused on solving very narrow problems with very narrow machine intelligence. Once a machine is trained to do something very narrow nearly as good as a human, it becomes very cheap to run that machine compared to running the human. Today, the technology doesn’t generalize well – the same machine that drives a car doesn’t adjust insurance claims. And that frustrates venture capital because they’re not getting their big exits. So, there’s a lot more investment chasing generalized intelligence. Since there’s an incentive to hype that effort, you’re hearing a lot of hype.
Practically every industry is, or will be, impacted by AI. What is one of the more interesting applications of AI happening in your industry / line of work?
CB: Deep fakes. It can very hard to assess the truthfulness of anything you read other people saying on social media. It’ll become increasingly difficult to trust audio/video.
JL: AI can manage production, distribution and inventory in a logistically-complex supply chain. By using vision and reasoning systems, we can make sure only the freshest produce is on shelves for consumers, with consistency, throughout the year.
PS: Hopper has added an interesting chapter to the story of the traditional travel agent. In older days, travel agents were skilled experts that helped you make travel decisions, but you paid a fee for the privilege. With the dot-com revolution, a bunch of companies created “self-service” websites where consumers could find and purchase their own travel products, and the Online Travel Agency (OTA) was born. Consumers could save money by forgoing a travel agent, but in turn, inherited the stress of shopping for hours trying to decide what to buy, and purchasing in frustration, unsure if they were purchasing at the right time or from the right website, and potentially over-paying for their trip. This left the question of whether saving a few dollars on a travel agent was really worth it. At Hopper, we are bringing back the expert adviser in the form of a mobile app: tell the bunny what you’re looking for, put your phone back in your pocket, and let our algorithms do the work to track and compare prices, ask questions and make suggestions to help you find the right trip (maybe even something you hadn’t originally considered).
On the other hand, AI also has the capacity to completely disrupt aspects of everyday life. Which application(s) of AI do you think will be most disruptive to the average person?
JL: AI will be a democratizing force across industries but has the greatest potential to upend one of the most fundamental aspects of our species: eating. For example, by aggregating and analyzing information on where food comes from, how it was made and how it was stored we can predict food safety problems before they happen.
Modern grocery and logistics is a process of managing complexity; the advent of big data and AI will make more products, of higher quality, available for less money and attuned for individual consumer tastes.
CB: How the state decides to apply the tool of narrow machine intelligence to governing society. There are so many laws on the books that you likely violate a dozen every day. As described in The Dictator’s Learning Curve, all that an authoritarian has to do is selectively enforce existing laws against its select opponents and claim its legitimacy through that action. Narrow machine intelligence is capable of scaling that activity and can add a rhetorical veneer of legitimacy. Sadly, that’s incredibly effective.