If you find yourself getting lost a lot less these days, you can thank Brian McClendon. In this edition of our C2FO 20/20 series, C2FO technical advisor, McClendon shares a look back at his experience as one of the creators of Google Earth and former VP of Google and shares what every business person needs to know about the future of self-driving cars, machine learning, and AI.
C2FO: So, Brian, can you share a bit about your career to date?
Brian McClendon: I am an engineer by training, I built computers for ten years, specifically graphics supercomputers — the fastest 3D graphics of their day — for a company called Silicon Graphics. These were used for things like movie animation and flight simulation. The experience got me focused on what you can do with 3D.
We did some demos for flight simulation that led to a 3D model of the world on this very expensive computer. A group of us were inspired to build Keyhole, and our product, Earth Viewer, a couple of years later. That was the product that became Google Earth.
Keyhole Earth Viewer became very popular on CNN in 2003 during the Iraq invasion. That led to Google’s interest in us and acquisition of Keyhole in 2004. I then led the team called the “Geo Group” at Google. We built Google Maps, Google Earth, Street View, Sketch Ups, and other products. I was there for ten years and grew the team from 29 people to 2000 people. I left the Google group to go to Uber and lead their maps and work on self-driving as well as their business platform group for two years before coming back home to Kansas.
C2FO: What have you done in your career that made the most difference?
Brian McClendon: I think it was in the release of Google Earth and Google Maps in 2005. We changed how people use maps and locations by making it fast and easy. Paper maps were slow. Online web maps before that were faster but still slow. With Google Maps, it was extremely fast; it was fluid. The difference in the impact you have if your product works at the speed at which humans think, instead of how slow-processing computers think, changes how people use products. That’s why Google Maps became popular, not for any other reason; it was fast. And Google Earth, in a similar manner for 3D, worked, as you could go wherever you wanted smoothly. People got used to this idea that a map of anything you wanted was available at any time.
One of the things that Google Maps found out early on was that people wanted to put their data on maps or use Google Maps for their purposes. The API allowed any other website to use Google Maps. Last I checked, there were 24 million different websites using Google Maps in some form. The power to put your own data on top of maps meant that people could tell stories or explain specific location-based facts and have the context of either the satellite imagery or the Maps’ data. People didn’t have to solve that problem. They could just use the Maps’ data to do it. I can’t call out one use case out of the 24 million, but there were a lot of very interesting use cases.
C2FO: An odd question for someone who has been on the leading edge of technology, but is there something you wish you would have learned sooner or done differently?
Brian McClendon: I think that the challenge with building any product is timing. We started working on what became Google Earth very early. In 1998 the number of people who had 3D graphics on their computer was very small. We couldn’t have made a business of it because if you told people to download it, and if they did, it would immediately not work. That was a problem. It wasn’t until 2003 and 2004 that there were enough computers out there that would run it.
You can do something that is great, but not ready for wide-scale adoption. The question is “what is your business until the world catches up with you?” One of the things we needed to learn — that I think any company needs to learn — is that if they are early, they need to figure out some business that’s ready to pay sooner. Or, figure out if they can offer something similar but not quite the same, for three or four years until the product is ready for consumer adoption.
These days if it runs on the phone, then you’re done. You can install it and run it anywhere. Back then, there was a very wide variety in what was possible on people’s computers. So, we sweated for a few years waiting for the world to catch up with our idea.
C2FO: What was your best career decision?
Brian McClendon: Selling to Google. And, it was a decision, because we weren’t sure about this company. It wasn’t IPO’d at the time, and we didn’t know what its finances were like. The claims they were making to us seemed highly unlikely. But those were completely correct, of course.
C2FO: You are leaving this leading edge of tech behind to embrace a new role in politics, what will you miss?
Brian McClendon: There are still opportunities to be involved with tech here in Kansas and to work with companies that are doing interesting things. Part of it is that I wanted to come home, and this was an opportunity that wasn’t available before.
C2FO: What should we know about the future for Google Maps and technology like AI and machine learning?
Brian McClendon: Part of the future will be around virtual and augmented versions of the real world. Google Earth’s 3D model of the earth is very powerful and could be central to making Google successful in those areas. There are parts of virtual and AR that don’t need a real-world model such as game playing and data modeling where everybody can compete pretty well. But Google has a strong lead in building a 3D model of the real world, and that’s going to be very useful.
AI is an “everything to everyone” kind of label. I would separate it out. Machine learning is a specific task-targeted subset of AI. It will make a big difference in the short term. You can apply machine learning to all kinds of problems and get better solutions. We are going to see five to ten years of machine learning just solving interesting problems more efficiently than humans have in the past, even solving some problems that had never been solved before because the signal is hidden in the noise, and machine learning managed to find it.
What most people think of as AI, artificial intelligence, is still some ways out. I don’t think computers are going to be teaching themselves things that are useful very quickly. Right now, the only success we’ve seen is where computers have learned to play games. They succeed, mostly, because they play themselves, and have a significant amount of time to play — whether it’s chess, or Go, or whatever. So, they’ve managed to create some interesting and powerful game programs because they can simulate against themselves until they figure it out. The real world is much harder because you can’t run 10,000 or 100,000 of these things in parallel and get better quickly; you have to collect data at the speed of humans.
C2FO: You’ve spent the last couple years at Uber’s maps division. What’s ahead for self-driving technology?
Brian McClendon: In general, I think that self-driving is going to be part of a car service. Uber and Lyft are preliminary versions of a long-term ride-sharing service where computers provide the driving and you just borrow a car when you need it. I think that change is going to be in ten to 20 years. The majority of Americans won’t own cars because it won’t make financial sense. It will be cheaper to borrow a car when you need it.
C2FO: Do you have any fears about the future and AI?
Brian McClendon: I think there are potential issues with it, but I don’t think these are close. I think we can solve many problems tactically before we face the challenge of what happens when a computer can learn by itself, in its own direction. First, a computer would need to be able to self-modify, and second, a computer must be able to build the next physical incarnation of itself — a robot building a robot. Those two things are necessary before there is something to truly fear. Neither of those things is possible right now. I can see it taking decades to get there. It’s something we need to figure out, and we don’t have a solution right now. There’s nothing we can do except solve the problem. You can’t stop us from heading that direction.
C2FO: What’s the best thing business leaders can do now to prepare for this “fourth industrial revolution” with machine learning, AI, robotic process automation?
Brian McClendon: To successfully apply machine learning, companies need to collect data on their operations and correctly label success/failure in their system. If they have enough data, machine learning can make good predictions about how to optimize their system.
C2FO: Any advice for the rest of us trying to understand this new future?
Brian McClendon: Enjoy the ride. Presume that things not solvable in the past will be in the future. For example, medicine and medical technology are going to jump by leaps and bounds. How we spend the minutes of our lives is going to change as much in the next 20 years as it did in the previous 20. Be ready for that change. Think back to what it was like in 1998 and compare to that to today. It will be at least that amount of change. Possibly more.
In addition to eyeing a future in Kansas government with a goal of protecting the voting system from hacking and using smarter technology approaches to improve efficiency and save costs, Brian McClendon is a technical advisor for C2FO, helping with machine learning, data science, and software-at-scale initiatives. He is currently a software designer, developer, engineer, advisor for Firebrand Ventures, and a research professor at the University of Kansas. Brian served on advisory boards for both the School of Engineering, the Department of Electrical Engineering and Computer Science, and serving as a Trustee for the KU Endowment. In 2007, he and his wife, Beth McClendon, established the McClendon Engineering Scholarship at the university.
The 20/20 series
C2FO was founded by taking a clear look at a broken finance system, and redefining how it could work better in the future. Our vision is defined, not by what we created, but by the customers we serve. From business leaders who manage the world’s largest companies to the entrepreneurs of all sizes who access working capital to fuel growth, and build the economy and the innovations of the future, our network is a humbling collaborative of thinkers, doers, and creative and economic forces. They help us see lessons of the past and goals clearly, and visualize what is possible ahead. Our 20/20 series focuses on the people who comprise the C2FO network. Through their stories, we reflect on goals and lessons and envision what is possible ahead.