Artificial Intelligence: A tool for bridging technology and management
[Part 1]How AI can change your business: Three gaps to bridge
Dialogue: Vijay Kumar (Dean, University of Pennsylvania) and Takayoshi Yamakawa (President, Dream Incubator)
Recent trends in artificial intelligence (AI), the “Internet of things” (IoT), robotics, and other new technologies are sparking transformations in the business world. To keep pace with these momentous changes, Japanese companies will need to make connections with cutting-edge human resources from around the world and work to make up ground on the technological vanguard.
Japanese companies need to bridge three gaps:
(1)Technology and management
(2)Japan and the United States (the international scene)
(3)The old economy and the new economy (in-house philosophy vs. collective intelligence)
To find out more about closing those gaps, Dream Incubator Inc. President Takayoshi Yamakawa sat down with Professor Vijay Kumar (Dean of the School of Engineering and Applied Science at the University of Pennsylvania), an international authority on AI, expert on drone technology, and driving force in industry-academia collaboration. This three-part series details their conversation, which revealed valuable insights on the topics of AI, technology, and business.
Session 1: Bridging the gap between technology and management
■Applying state-of-the-art technology in the business world
Yamakawa: AI development seems to be surging on advances in other areas, like increases in computing power and the evolution of network environments. How do you think those different sectors impact AI?
Kumar: When you think about AI, you have to be aware of the two technological trends behind it. The first big movement centers on computing power, which continues to double on a yearly basis. Moore’s Law might be coming to an end, but if you look at projections for the next 5 years, people are still predicting computers to be about 30 times more powerful than they are today.
The second trend is the explosive growth of data. Over the 10-year span between 2010 and 2020, the data volume is going to have grown 50-fold. Companies are trying to sift through all that data, piece together elements they can use, and leverage what they find in revolutionizing their businesses. Things like AI and deep learning are right at the center of the changes.
Yamakawa: One of the things I want to talk about is how businesses are supposed to apply the new technology trends.
From my point of view, businesspeople don’t always know that much about the technological aspects of computing and data. The same thing happens on the other side, too: Researchers don’t always understand how technology can have its biggest global impact through business channels, either. There always seems to be a big gap between technological advances and business applications.
■AI can only provide you with probability
Kumar: Think about what businesspeople need to know when they’re trying to use data to make decisions. To me, they need to have a fundamental understanding of what probability is, what AI is, and what data is.
The first key thing is that AI can only give you probabilities and predictions based on data—it’s your job to use that data and those probabilities to figure out the causal relationships between events, formulate a plan, and take action.
Yamakawa: You’re saying that AI doesn’t give you “answers”—it can only provide “probabilities.” That’s where people come in, using accurate data and probabilities to model the mechanisms defining how things actually work. Without that human element in place, there’s no way to make informed decisions.
Kumar: That’s correct. AI is just a tool.
Yamakawa: That’s a really interesting insight. People have traditionally based their decisions on know-how and past experience, but there’s an ever-growing pool of data and accurate probabilities at our disposal now. How do you think companies are supposed to bridge the gap between accumulating those huge reservoirs of data and actually making decisions with those resources?
Kumar: First of all, there’s no substitute for human knowledge. You have to model that intelligence to the best of your ability and then combine what you can with data-driven approaches.
There’s only one basis for a data-driven approach: data. On the other hand, people have been creating “models” in so many different fields and using them to explain things. Robotics uses models from physics, for example.
Data-driven approaches are new and exciting, which makes model-based approaches seem old, traditional, and boring. That said, I still believe that both sides—data and models—have to come together.
We can’t afford to stop learning. To me, the key is understanding how to get from data to information to decisions.
■Tailoring AI to distinctive business elements
Yamakawa: Could you share any examples of how businesses are applying AI and deep learning?
Kumar: How about the entertainment industry? There’s a company called Anki Drive, which makes a racing game where you control your toy car with an iPhone—like it’s a remote control. Your opponent is an autonomous car with built-in AI.
That AI car learns the optimal driving techniques through deep learning: It collects and analyzes huge amounts of operating data from other toy-car drivers all over the world. Remember—this is a toy we’re talking about. It’s a cool example of how deep learning can enhance entertainment value.
Yamakawa: It’s all about fusing human know-how from a wide array of sources and then augmenting it through deep learning, which creates AI that’s capable of “beating” a human. That’s pretty amazing.
Simple data can still have a big impact, too. Just in the entertainment industry, even, there are ways for companies to harness data into something powerful without having to go as far as high-level deep learning.
Kumar: That’s true.
Yamakawa: Artists and promoters would be able to explore so many business opportunities if they had data on their fans and customers. Let’s say a popular singer is doing a concert in Las Vegas. When people start buying e-tickets to the show, organizers will not only be able to know who’s going to be sitting where but also be able to predict who’s going to be coming, when they’re going to arrive, where they’re going to be coming from, and where they’re going to be heading. They’ll even know what kinds of concerts people have gone to in the past and what kinds of souvenirs concertgoers have bought at the shows. Thanks to data, promoters could theoretically even offer to make advance plane or hotel reservations for guests. The monetization possibilities are endless.
In the past, all that information came from know-how. When computing and data analysis technology enter the picture, I think the new insights will enhance the whole process of going to a show—for artists and fans alike.
■What it takes to connect technology and management
Yamakawa: Here’s one question I want to ask you. In Japan, the director of a company’s research and development (R&D) division often serves as the company’s chief technology officer (CTO)—but CTOs don’t necessarily have any business experience. Do you have any advice for Japanese companies on that point?
Kumar: In the United States, too, you often see CTOs with lots and lots of technical background but not necessarily any management experience. To survive in this world, though, a CTO has to combine technical smarts and management skills.
Yamakawa: I heard you spent half a year at the Tokyo Institute of Technology back in 2003. What do you think of Japan?
Kumar: I don’t know much about Japanese organizations, but my superficial impression is that they’re quite hierarchical and vertical; there’s not enough cross-talk across departmental lines. That makes it harder to learn about things outside your specific area of expertise.
At the Pennovation Center, which the University of Pennsylvania created to promote open innovation, there are no private offices—not for students, researchers, or even me. That’s the kind of environment that helps scholars and companies from different fields team up on research projects—and makes it easier for people to glean knowledge from outside their domains, giving them the opportunity to broaden their horizons. I’ll talk more about the Pennovation Center in our second session.
Yamakawa: Like you were suggesting, people working for Japanese companies tend to assume that following the rules of their respective industries or departments is the quickest path to success. The problem with that mindset, though, is that you can’t rely on the people around you for all your information. Your department, your company, and your industry aren’t the only places where technological development is happening—top-flight engineers from all around the world are pushing technology forward on a daily basis. CTOs in this day and age have to take a broader, more global perspective on technology, I think.
Kumar: You’re exactly right. I don’t think you can afford to learn just one thing—regardless of what environment you’re in. Moving into the future, I think people will have to know lots of different things at some level instead of knowing one thing in great depth.