A→I Artificial Intelligence: From "Artificial" to "Intelligent" (Part 1)

Some view artificial intelligence pessimistically as "something that will take away jobs humans do," but fundamentally, it should be used to expand the ways humans live. This Dentsu Inc. Design Talk explores the future brought by artificial intelligence. The speakers are Dr. Yoshiki Ishikawa, a medical doctor researching preventive medicine × AI; IT entrepreneur Dominic Chen, also a leading figure in information science research; creator Tetsuya Mizuguchi, who constantly incorporates cutting-edge digital technology to create new forms of expression; and Hiroshi Yamakawa, Director of Dwango AI Research Institute. They discussed a future where AI is not just 'Al' (artificial) but naturally 'Il' (real), focusing on creativity, desire, the heart, and emotions, viewing technological progress positively. The session was moderated by Fumi Hito of Dentsu Inc. We present the discussion in two parts.

(From left) Mr. Mizuguchi, Mr. Ishikawa, Mr. Dominique, Mr. Yamakawa, Mr. Hito

 

AI has an impact comparable to the Industrial Revolution

Nitta: My theme is the future-oriented industrial application of accelerating technologies like artificial intelligence. While AI is currently in its third boom, it has evolved from handling only numbers and text to now processing unstructured data like images and audio. Image recognition is said to have surpassed human accuracy, and speech recognition is also advancing rapidly. What exactly is AI, and how does it connect to the next wave of business? That is the purpose of our discussion today. Recently, macroeconomists have begun referring to AI. Professor Tomohiro Inoue of Komazawa University states that AI is a "general-purpose technology." A general-purpose technology is one that can be applied across various industries. Steam and electricity, which drove past industrial revolutions, were also general-purpose technologies. In other words, history has progressed along the line "steam → electricity → IT ≈ AI," and the argument is that AI can drive economic growth.

Yamakawa: I spent about 22 years researching artificial intelligence and related fields at Fujitsu Laboratories, then moved to Dwango AI Research Institute in 2014. From around 2007 to 2010, I was involved in research on the brain activity of professional shogi players when intuitively selecting their next move. Back then, while computers could analyze hundreds of millions of possible moves, they were poor at judging the quality of the next move. Humans, on the other hand, could only see one or two tactical patterns, but their intuition for recognizing those patterns was powerful. The machine vs. human match was essentially a contest of "calculation" versus "intuition." However, recently, deep learning has enabled AI to develop a certain level of intuition. As a result, the power balance has shifted, and humans have started losing.

Now, what about creativity? What is creativity, fundamentally? Its core lies in breaking down a whole into parts and then combining them. Deep learning has enabled AI to extract components like eyes, noses, and mouths from facial images. And AI has always excelled at combining elements. However, in many cases today, selecting high-value combinations remains beyond automation. In terms of creativity, current AI has not necessarily surpassed humans.

 

What is the "creativity" AI demonstrates?

Dominique: I'm not a specialist in artificial intelligence or machine learning, but for about ten years, I've been involved in activities concerning the nature of copyright in the internet age. Current copyright law doesn't fit the online environment where creativity is expressed through modification and remixing. Therefore, I've worked to promote "Creative Commons" as a system where creators voluntarily express their intentions regarding the rights to their works.

I believe I was invited here today because I co-edited the Japanese translation of Singularity: From Artificial Intelligence to Superintelligence, published in January. This book calmly and matter-of-factly discusses the differences between human intelligence and machine intelligence. Reading it makes clear that AI is no longer a distant future concept. As seen in its application to collaborative filtering and recommendation engines, it is already beginning to permeate society as a whole. When considering the theme of AI and creativity, what remains for humans are non-quantifiable preferences and sensations, along with the innate motivators of life—pain and desire. I believe the most fascinating question now is how to handle these as information technology within AI and VR.

Ishikawa: I research preventive medicine, but I ended up studying AI as well through a chance encounter. Initially, I was researching diets. Then I discovered a fact: over 80% of people who lose weight through dieting from spring to summer regain their original weight or more by summer or after New Year's. In other words, dieting was actually a "surefire way to gain weight." Therefore, to avoid gaining weight, you must improve your diet before you start gaining. How can we create dishes that are both delicious and healthy? As my research progressed, I realized that if we input three elements into AI—recipes, ingredient flavors, and cooking methods—it could create new, tasty dishes. Right around that time, I learned that Rav Varshney, a genius scientist of my generation, had already created Chef Watson. So now, I'm collaborating with him on this research.

In the very near future, AI capable of creating healthy, delicious meals will emerge, ushering in an era where every household has a Chef Robot. These Chef Robots will deeply understand individual preferences, knowing better than the person themselves what they want to eat and when. Moms are like that, right? "It's about time for curry, isn't it?" (laughs). Once this becomes reality, both ingredients and beverages could be automatically procured via drones. In other words, health might become an "infrastructure" service, just like gas or water.

I also research creativity. The root of "creativity" is "nature." Humans pose questions, gather diverse information, combine it, and generate ideas. But nature doesn't pose questions. Moreover, DNA communicates using just four letters: "A," "T," "G," and "C." This makes its creative speed vastly faster than human language. Nature creates things that are statistically improbable. Therefore, if humans can create something statistically improbable, that would be the first time we could truly call it creative. Personally, I believe that if I could combine ingredients in a statistically improbable way to create a divine dish, I would have achieved creativity in cooking. How humans and machines collaborate to achieve this is, I think, our challenge in the 21st century.

Mizuguchi: I'm a game creator. Last December, for a WIRED magazine project, I experienced Audi's AI "Robby" racer program at the Barcelona circuit. When I got in the car, it immediately said, "Welcome, Mr. Mizuguchi," and it really hit me. An engineer sat in the driver's seat, but he only pressed buttons. After that, the car drove itself, reaching speeds exceeding 250 km/h on straights.

It was scary at first, but since the entire car is packed with sensors, it learns the road conditions over two or three laps and drives at the fastest possible speed with minimal loss. It never slides in corners and consistently posts the best lap times. Gradually, my feelings shifted from "scary" to "fun." On the way back, an Audi representative drove me to the hotel. When changing lanes on the highway, I caught myself thinking, "Isn't that a bit rough? Drive more smoothly!" (laughs). Just one ride in the lobby made me realize my own driving sense had been dramatically upgraded. That feeling remains with me now. I came to understand firsthand that the risk of human error is always present as the price we pay for the joy of driving.
What exactly is a car? From a media design perspective, all media exist as extensions of human bodily functions and senses. Cars were extensions of our "legs," but self-driving cars are now integrating the "brain" into that extension.

Mizuguchi: There's another book that became a catalyst for me becoming a game creator: 'Society of Mind'. It objectively observes humans, exploring what conflict is and how human emotions arise. In 2011, I had the chance to interview its author, Marvin Minsky, an AI researcher. His very first words shocked me. "Humans have free time, right? So they want to do something." As various tasks we perform get replaced by AI, humans will have even more free time. The question is: how do we transform that free time into something enjoyable? I realized that's the challenge facing creativity today.

<Continued in Part 2>

You can also read the full interview here on AdTie!

Planning & Production: Aki Kanahara, Dentsu Inc. Event & Space Design Bureau