The 100,000 meal problem solved by AI. The key to making life better for humans is cooking.

Yoshiki Ishikawa

Ajinomoto Co., Inc. newly established the Consumer Data Analysis & Business Creation Dept. in the spring of 2018. Its mission is to make the future of food enjoyable. In this series of FUTURE FOOD TALK, we shall look together at the different paths we should be taking by talking to people in different fields about the future of food.

If a person lives for 100 years, he will eat about 100,000 meals in his lifetime. How can we make those 100,000 meals better? Preventive medicine scientist Yoshiki Ishikawa is trying to improve the well-being of humanity through food by developing Food Galaxy, a program that visualizes and develops our food styles.

How does the AI program, Food Galaxy, come up with recipes?

Mr. Ishikawa: We’re working on Food Galaxy which is a project that can tell us where we are in the food galaxy by mapping recipes from around the world and putting them all on a single map.

Most recipes in the world are made up of about 10 ingredients on average which is less than 1% of all the combinations of ingredients on the planet. This means that 99% of all food combinations haven’t been tried. In other words, there’s hardly any innovation happening. Or to put it another way, we should be able to innovate by putting together different patterns. So we’re developing Food Galaxy based on these ideas.

It came to me when I was eating a California roll (sushi). I was wondering where’s this recipe from? Like, what percentage of the California Roll is Japanese? Equally, I was thinking where does this new idea come from if it’s not from Japan? That’s how I got thinking about the possibilities of combining different foods when I was getting together recipes from around the world.

The Food Galaxy methodology began with the flavor pairing theory. The theory is that different ingredients are more likely to pair well in a recipe when they share flavor compounds.

Western researchers who discovered this theory say (without going in to too much detail) that food is mostly determined by aroma. For example, if you drink orange juice and apple juice while holding your nose, you won’t notice the difference in taste. Starbucks has come up with different recipes based on this flavor pairing.

In the early 2010s, a paper was published saying the theory doesn’t work for East Asian cuisine. Surprisingly, there are more combinations of compounds that don’t have a common flavor, and Western researchers were struck by the fact that East Asian people don’t care so much about flavor.

So we looked at umami. Umami is like a food booster in that it makes the taste more pronounced but in contrast to Western food which includes one umami compound, we found that East Asian foodstuffs like konbu and bonito, vegetables and meat produce synergies by combining different umami flavors.

Aromas are made stronger by drying, and in western climates where it is dry, the aroma of food is important. But on the other hand, because East Asia is wet and humid, it suggests that umami works better on the tongue than aroma. We called this the umami pairing theory.

Although it is quite a bold hypothesis, it has become apparent that the world's cuisines can be classified into three patterns that focus on Western flavors, East Asian umami, and spices that are typical of India.

To make a dish that tastes good, the approach is quite different depending on the country or region. If you know the theory it shouldn’t be too difficult to customize your own dishes.

So we developed what we call a style transfer tool to utilize this idea. It’s an algorithm that allows you to freely change the style of a dish, like changing sukiyaki, a Japanese hot pot of meat and vegetables, into a German or French style.

Food Galaxy also has a function which we call Food Galaxy GPS. For example, if you enter your eating habits for the past 2 weeks or so, the system can identify what your own preferred tastes and food styles are.

To get to a new place, you’ve got to know where you are first. By knowing your current location in the food galaxy using this GPS, you can go to food destinations further afield.

If you really want to you can go back in time to the Japanese food of the 1970s which is said to be the most healthy, back from the 1980s onwards where the Japanese diet relies too much on junk food. I think style transfer is a way to explore these routes.

How to make the 100,000 meals we eat in our lives better.

Why am I doing this Food Galaxy research? The reason is because of what I call the 100,000 meal problem.

Assuming a person eats three meals a day and lives to the age of 100, the average person has about 100,000 meals in their lifetime. But if you count all the ingredients that exist in the world, they don’t fit in to the 100,000 meals. If that’s the case then we can’t rely on humans to make our eating lives any better. But AI could tell us what foods work best to make these 100,000 meals better. It’d be pretty cool if there was a world like that, right?

There are two major approaches to finding this path of better food. One is how close AI can get to the idea of a professional chef. The other is forgetting the history of ingredient combinations that humans have come up with so far, and how AI can independently evolve our food for us.

Through these attempts I predicted what kind of cuisine there’d be in the future Michelin Guide for 2100. It’s a good idea to factor in population changes, the result is that dishes in the 2100 Michelin Guide would use a lot of African and Chinese ingredients. I had a professional chef make one of these dishes for me to eat. But it was pretty tasteless (laughs). That said, someone who stayed in Africa for quite some time described the food as really tasty and said it brought back memories and felt kind of new, at the same time.

The key to people living better is cooking

Mr. Sato: Thanks. That’s all really interesting. I think what Mr. Ishikawa is researching is what it means to live better. Is there anything you look at from the perspective of food and well-being?

Mr. Ishikawa: I think that it’s actually cooking that makes better lives for people. When people’s use of time is divided into negative, neutral, and positive activities, research has shown that in place of negative activities (like work and household chores) which we spend less time on because of decades of advances in technology, instead of doing meaningful, positive activities (like learning and hobbies), we spend more time on neutral activities like idly watching TV or looking at our mobile phones. Even when the world becomes a more convenient place, we humans get bored. I think that cooking is the best way to ease this boredom.

Mr. Sato: You have to really think about how to turn your neutral time into something positive. I think the most important thing is time, but in order for food to make our lives better, do we need to cook quickly so that we can save time, or should we be creating a more valuable time with the communication we get from enjoying our food more slowly? I think it’s really difficult.

Mr. Ishikawa. The future of food and the use of time can’t be separated. One study which looked at the changes in calories consumed by Americans over the past few decades, showed surprising findings because the calories consumed for breakfast, lunch, and dinner hasn’t changed at all. What has increased (especially after the 1980s) is snacking, which has resulted in more obesity.

The evolution of meal kits, fast food, women’s empowerment, and people’s loneliness ... It’s because of these things that it’s said the time for cooking and eating with others is much less than it used to be. Looking at the relationship between the time people spend cooking and obesity around the world, obesity rates are lower in countries where cooking times are longer.

I think cooking is what enriches our lives, I think it goes beyond food, helping to create a person's style. What kind of clothing, food and living styles we have is important which is why I want Ajinomoto to create a style ingredient that merits spending time.

Mr. Sato: A style ingredient?

Mr. Ishikawa: I think there are lots of clues. One of the things I’m concerned about is a world where humanity lives in a hotel (laughs). It’s been said that 80% of humanity will live crowded into a small box, known as a city. The most important issue we face is how to make our lives better while we live and move around in such a small box. I think food will play a really big part in this.

Mr. Sato: Are there any examples of companies doing this?

Mr. Ishikawa: One of the most popular overseas health support programs is a cooking program with Harvard University. Top chefs teach you how to cook, you learn about nutrition, and you can start cooking from scratch. At the end of the course, participants entertain the people starting the next course by making food for them.

Mr. Sato: That must be quite popular? I’m really interested in it.

Mr. Ishikawa: Even in American hospitals, there are now more and more kitchens. For example, it’s more effective and more powerful to teach people how to shop and cook in hospitals than it is to give nutritional guidance to people with diabetes. Cooking is a compulsory class at the Stanford University School of Medicine.

Mr. Sato: The definition of food simply as that which is called food is a bit narrow, isn’t it? We should use food to create certain life styles. Our role in the future might be in some cases to provide things that aren’t food at all.

Yoshiki Ishikawa
Yoshiki earned a bachelor’s degree in Health Science from the University of Tokyo, a Master of Science in Health Policy and Management from Harvard School of Public Health, and a PhD in Medicine from Jichi Medical School.
As a public health researcher, entrepreneur, and science journalist, Yoshiki is working at the intersection of science, business, and government as a catalyst with the aim of advancing the well-being of society.
Ken Sato
Ajinomoto Co., Inc. General Manager, New Business Creation Group
  • LINE

Other Talk

Page top