AIバブルへの警鐘――人工知能への過剰な期待は禁物、今のAIに出来る事と出来ないことを見極めることがAIビジネスの成否を分ける

近年、AI（人工知能）の発達と実用化には目覚ましいものがある。たとえば日々、認識精度を上げているGoogleの音声検索、交通・輸送手段などに革命をもたらすと見られる自動運転車、さらにはコンピュータによる同時通訳の自動化など、AIが私達の社会や産業界に与えるインパクトは広く、そして深い。
が、その一方で最近はAIの能力が過度に喧伝された結果、それに対する期待が真の実力を遥かに上回るところまで達し、一種バブル化の様相を呈しつつある。
AIはその長い歴史において、過去に2度、現在と同じようなブームを経験した。そして、それがバブル化して弾けた後には、「AIの冬」と呼ばれる長い低迷期が続いた。その時期には投資が冷え込んだだけでなく、研究者がこの分野に見切りをつけて去ってしまうなど、長い目で見ればマイナスに作用した。当時と同じ過ちを繰り返さないためには、どうしたらいいのか？本レポートでは、現代AIの実力を冷静に評価することによってバブル化を防ぎ、これをもってAIの継続的な研究開発と産業化を促す一助としたい。

The recent progress and commercialization of A.I. technologies has been quite remarkable. For example, Google’s voice search is constantly improving its accuracy, autonomous cars are expected to revolutionize our traffic and transportation systems, advances in the computerization of simultaneous interpretation are commonplace, and many more technologies are constantly appearing. Undoubtedly, these numerous applications of A.I. technologies will have a profound and broad impact on our society and a wide variety of industries in the near future.

On the other hand, it is also possible to see the hype surrounding A.I. is clearly surpassing its true capabilities, which this article suggests is creating a sort of A.I. bubble. A.I. has already experienced two large booms over its long history. Unfortunately, unfounded exuberance meant these periods of rising expectations both turned to bubbles, which inevitably burst several years later, and lead to very long periods of stagnation, periods of so-called “A.I. Winter.” During the dark periods, not only did investment in A.I. technologies drop significantly, to the point of practically disappearing, but also experts of A.I. gave up and left the field one after another, causing long-term damage to the research and development of the A.I. sector as a whole. This current period of growth can be considered a Third Spring for A.I., but we should also be conscious of what can and should be done so that we do not repeat the same mistakes made in the previous eras of rapid advance. In this report, we would like to evaluate and examine the true capabilities of current A.I. technologies in order to prevent the formation of a potential 3rd A.I. bubble, which, in turn, will help facilitate the development and industrialization of A.I..