AI, Public Policy, and Aging Society, with Yoshinori Hiroi
2018 Uehiro-Carnegie-Oxford Conference: Ethics and the Future of Artificial Intelligence
May 17, 2018
YOSHINORI HIROI: Hello, everyone. My name is Yoshinori Hiroi—just call me Yoshi—from Kyoto. Kyoto is, as you may know, a traditional sightseeing town, but as I listened to Arisa's talk, maybe there are going to be a lot of robot hotels in Kyoto. So you may stay at the robot hotels.
I have been doing research in the area of public policy and its philosophical foundation, and so I will talk about AI (artificial intelligence) and public policy and a little bit about aging society.
First, let me show you just an example about what AI can do. Incidentally, last year I did collaborative research with Hitachi—Hitachi is an electronics company, like IBM, in Japan—at the Hitachi Kyoto University Laboratory on the Future Sustainability of Japan Using AI Technology.
We made a cause-and-effect model consisting of about 150 elements, like this kind of model, and made 20,000 simulations regarding the future state of Japan in the year 2050 from the perspective of various aspects of sustainability—like demographic sustainability, financial sustainability, and so on.
This is the model, AI simulations. This is a snapshot of a chronological trend of the changing future of Japan—this is the year 2042—20,000 variations or scenarios. This is a rather simplified picture. We assess it from the perspective of sustainability. Our AI-based simulations reveal that a decentralized and localization model will become desirable from the perspective of sustainability—health, equality, and happiness of the people.
Although this research is still very primitive, it shows an example that AI can broaden our future perspective or vision and contribute to the various decision-makings, including public policy decision makings.
I talked about positive aspects or possibilities of AI, but actually AI is far from perfect and we should think about the basic perspective of what AI can do and what AI cannot do.
I think many of you here are already familiar with this evolutionary model of the human brain by Paul MacLean.1 According to this model, the human brain consists of three parts reflecting the process of evolution.
The first and oldest part is basal ganglia, or brain stem, which deals with instincts, drive for survival, that kind of thing. The second part is the limbic system. This is developed in the mammals. This is in charge of emotions or social communications. The third part, neocortex, is in charge of thinking or reasoning. This is highly developed in the human beings.
Just simply speaking, what AI can do or what AI is good at is this part, thinking or reasoning. What AI cannot do or AI is not good at is instincts or emotions or social communication. This in some sense relates to the morning session's discussion. In my understanding, AI is a machine that has only the third function of the brain, set apart from the first and second parts.
But the important point here is that this third part is based upon the first and second parts, so it cannot be an independent function. So, in principle, although AI has the ability of much faster speed of reasoning and bigger amounts of memory, it does not have the ability of emotional interpretations or understanding the meaning of the words.
In this point AI is nothing more than a tool for the human beings. But although AI is nothing more than a tool for the human beings, as in Arisa's discussion, AI can replace a certain portion of human labor or employment. In this context, recently, as you have already heard, there have been a lot of discussions about the necessity of so-called basic income (BI) or universal basic income, a social system providing a certain amount of money or income to all citizens, as most of the people will lose jobs because of AI.
What should we think about this kind of discussion or the necessity of basic income? I have been doing research about the welfare state policy. This is the evolution of capitalism and the welfare state. If we look at the historical evolution of capitalism or the welfare state, we can find a historical evolution from salvations—I mean kind of curative measures—after the incidence to earlier or preventive interventions.
From left to right—do you know who she is? Maybe from people from the United Kingdom know. Elizabeth I is famous for Poor Law in the early 17th century. This was the time of the rise of the market economy or capitalism. That led to the Poor Law.
The central person is Bismarck in the 19th century. The Industrial Revolution occurred, more people were at risk, and then they had social insurance in the 1880s. This is, in a sense, more kind of preventive measures than the Poor Law.
But again, the Great Depression occurred in 1929 in the 20th century, and a huge amount of unemployment.
But after the latter half of the 20th century we had the so-called Lehman shock and also AI technology—there has been a lot of discussion that they lead to more unemployment and the necessity for basic income also.
It is possible to think that we need preventative measures against unemployment and poverty. In the ultimate sense, that is basic income. In this context basic income may be seen as a social system in the final stage of development of capitalism. But, as for myself, I am rather skeptical about the view that AI will replace most of the human labor and employment as the capabilities of AI are rather limited.
This leads to the topic of AI and aging society.
This is the front page of The Economist in November , "Japan's Burden." Another key word here is "Japan Syndrome." That means that the [nature] of the problem that Japan is facing is aging and population decline. But also, this feature article says that many other countries are experiencing the same pattern as Japan, aging and population decline, so this is common in a sense to all industrialized countries.
You see that this is the ratio of the elderly people, the historical chronological change. Japan is very high in comparison with the United Kingdom and the United States, the highest among the industrialized countries.
Also, if you look at this picture, the vertical axis is average life expectancy and the horizontal axis is the health care expenditures against GDP. Currently, Japan's performance is relatively good, relatively lower health care expenditures and relatively higher average life expectancy. So far, so good. But from now Japan is facing a still higher rate of aging.
The issues of aging are a common agenda for the industrialized countries, as I said, including the United States, United Kingdom, and Japan. AI technology can be beneficial in terms of caring or conversation robots or tools for frail elderly people, people with dementia, or care providers.
This is again from the Hitachi picture, communication and assessment robots for frail elderly people, like I ordered this yesterday online, this kind of conversation. So AI technology can be beneficial in that kind of sense, or complementary workforce for the situation of declining labor force, and so on. But the potentials of AI in these areas remain to be seen. The relationship between AI and aging society may be an interesting topic.
Thank you very much.