Japan in Their Own Words (JITOW)/日本からの意見

How can AI and humankind coexist?
ONO Goro / Emeritus Professor, Saitama University

November 15, 2018
Ever since one machine after another emerged to replace manpower in the wake of the Industrial Revolution, people have been more plagued by the fear of losing their job than comforted by the expectation of being freed from the pain of work. AI is thus feared to deprive some workers of their jobs in the short term. It is argued, however, that AI must benefit humankind in the long term.

AI may well have utility in making up for the imperfections of humankind in such areas as thorough checking of facts and developing logical argumentation. It may also be useful in preventing conflicts in laws and regulations and also miscarriages of justice. That said, it would seem too much to expect AI, which has no organic life, to step into and pass judgments on conflicts arising from differences in values held by sovereign states or nations, or by individuals.

The United States chose, for reasons of objectivity, the value or metrics of money as the common unifying value as it strove to preserve its unity while, as the ethnic melting pot, embracing diverse values. That is why Americans took the lead, before anybody else, in giving priority to the use of AI.

It should not be forgotten that AI has what machines and robots in the past did not have, that is, the capacity to undergo evolutionary changes. Would it be possible, then, that AI, freed from the constraints imposed by humankind, would attack humankind with an overwhelming force? This question has prompted some scientists to propose that a certain set of rules should be laid down to forestall such a danger. However, there is no guarantee that nobody would try to break such rules. It would be natural to assume that there would inevitably be those who would think of some ruse to get around them.

Further, how would you define “benefiting humankind”? For example, almost everybody detests poisonous snakes. But they sometimes play a part in exterminating vermins such as rats and mice that raid grain crops. An ecosystem can only survive only if there is diversity. If a certain element of the system is removed for one reason or another, some adverse consequences will inevitably follow.

--- Such would be the future of AI as seen by humankind. Then, what would the future be like if seen by AI, as the moment of singularity (a hypothetical future point in time when artificial intelligence will surpass human intelligence) fast approaches? To put it bluntly, would AI be inclined to “coexist with humankind” or not?

Seen from AI’s point of view, it can replace human-beings in a number of areas; not just in simple tasks that robots can perform, but also in most of the work done by pundits known for their expertise and IT engineers who are highly thought of in human society. Because, as far as mere accumulation of knowledge and deduction therefrom are concerned, AI is superior to humankind.

You might say “But humankind has creativity that AI does not have”. However, AI has exhibited far more creativity than human champions in the world of games. And it is not limited to narrow areas. Cloud data permit AI to have access to and use far greater amount of information than is available to humankind. AI would be far superior to humankind because of their capacity to produce something by analyzing such data, making deductions, and combining all these things. In this sense, AI would not be in need of humankind. In contrast, AI is powerless in those areas or concepts where there are no data or precedent, and would need those individuals who can create something that has never existed before. However, even if you looked for them all over the world, you would find at most only a few individuals who have such an extraordinary talent, not enough by a long shot to sustain the survival of humankind.

What kind of quality does humankind need to coexist with AI? Assuming that it lies outside the areas where AI excels, it must be the polar opposite of the kind of qualities that are known to be AI’s intrinsic strength and have been hailed in contemporary society, such as “scientific”, “logical”and “digital”. In short, it would be “ambiguity”. “Ambiguity” here does not mean something that can be expressed in some mathematical formulation. Rather, it means “ambiguity” in the existential sense that neither quantum theory nor logic can exclude. This may appear similar to, but is totally different from some concepts resembling ambiguity that have appeared in the past through, for example, the use of mathematical formulae.

Ironically, AI in the future would not really need AI technical experts, whose logic and approach would be IT-based. What it would need would be a set of ordinary, seemingly mediocre people. They would represent chaotic diversity, and chaos would be the factor that AI would need to make up for its own lack of diversity.

It should be noted, however, that the term “diversity” here is used not in the recent sense of the word, that is, according to its Western definition. To characterize something as the sole universal truth according to Western values would be tantamount to “losing diversity”. The diversity that we seek would lie in our “mutually recognizing differences”.

In other words, the time has come for humankind to effect a bold paradigm shift from binary choice propagated as a universal value by the West – based on the binary number system best suited to AI – to “diversity” and “ambiguity”.

Goro Ono is Emeritus Professor of Saitama University.
The English-Speaking Union of Japan

小野 五郎 / 埼玉大学名誉教授

2018年 11月 15日


しかし、AIには過去の機械類・ロボットには無かった自らを進化させる力があることを忘れてはならない。 そんな人類からの制約から離れたAIが人類に対し圧倒的な力で攻撃してくるようにならないか?との疑問から、科学者の中には「そうならないよう予め一定のルールを設けておこう」という提案も出されている。しかし、誰一人としてそのルールを破らないという保証はない。むしろ必ずや抜け駆けしようとする者が出て来ると考える方が自然である。

まして、何をもって「人類のためにプラス」とするのか? 例えば、ほとんどの人が忌み嫌う毒蛇にしても時に穀物を荒らすネズミなどの駆除に貢献するように、多様性があってはじめて生態系は成立するのであり、そこに存する特定の何かを除去すれば必ず何らかの副作用を伴う。

・・・ここまでは人類の側から見たAIの未来である。では逆にシンギュラリティ(AIが全人類の知性を超える未来のある時点)が間近に迫ったAIの側から見たらどうか? 直截に言えば、AIの方が「人類と共生してもいい」と考えるかどうか?である。


では、AIと共生する上で、人類に求められる能力とは何か? それはAIが得意とする分野の外にあると考えれば、AIが最も得意とし現代社会がこれまで重視してきた「科学性」「論理性」「デジタル性」とは真逆なもの、要するに「あいまいさ」である。なお、ここで言う「あいまいさ」は、数式によって表現されうるものではなく、あくまで量子論や論理学上も排除しえないような実在としての「あいまいさ」そのものである。 これは、過去に現われた幾つもの類似概念と全く似て非なるものと解されたい。


一般社団法人 日本英語交流連盟