ロボットに「できる仕事」と「採算が合う仕事」は違う

5/10/2026

2026年9月30日付のAnthropicの研究報告「What work can robots do? ロボットはどのような仕事ができるのか？」（Russell Legate-Yang、Maxim Massenkoff）によると、米国の全労働時間の約34％（物理的作業の約74%）に相当する作業は、一定の条件下で現在のロボットが実行できる一方、人間より低い費用で実行できる範囲は約0.3％と推計されており、「技術的にできる」ことと「採算が合う」ことは別であることを示しています。
そして、採算が合う範囲が10％に達するまで「約40年」と試算しており、日本の勝ち筋を考えるうえでは、「優れたロボットをつくる力」に加え、「現場で採算を成立させる力」に注目すべきでしょう。知財活動でも、性能向上だけでなく、導入・調整の手間や運用費用を減らす発明、現場データやノウハウの蓄積・活用に目を向ける必要がありそうです。当然のことですが、「できる」と「採算が合う」の間にこそ、フィジカルAIの技術課題と事業機会があるのではないかと思いました。
生成AIに、Anthropicの研究報告「What work can robots do?」及び関連情報を調べ、日本の勝ち筋について深堀させましたので、ご参照ください。なお、生成AIによる調査・分析結果は、公開された情報だけに基づくものであり、必ずしも実情を示したものではなく、誤った情報を含む可能性があることにご留意のうえ、ご参照ください。

What work can robots do?
Sep 30, 2026
https://www.anthropic.com/research/what-work-can-robots-do

2026年10月01日
ロボットは人間の肉体労働の74％をすでに実行可能だが「人間より安く働ける仕事」に限るとわずか0.3％という調査結果
https://gigazine.net/news/20261001-robot-job/

What Robots Can Do and What They Can Do Profitably Are Two Different Things
According to Anthropic’s research report “What work can robots do?” by Russell Legate-Yang and Maxim Massenkoff, dated September 30, 2026, today’s robots can, under certain conditions, perform tasks accounting for approximately 34% of all working hours in the United States (about 74% of the time spent on physical tasks). However, the share of work they can perform at a lower cost than humans is estimated at only about 0.3%. These findings highlight the distinction between what is “technically feasible” and what is “economically viable.”
The report also estimates that it would take “about 40 years” for the economically viable share to reach 10%. When considering how Japan can gain a competitive edge, we should therefore focus not only on “the ability to build advanced robots” but also on “the ability to make their deployment economically viable in real-world settings.” Intellectual property activities, too, may need to look beyond performance improvements and give greater attention to inventions that reduce the effort required for deployment and adjustment or lower operating costs, as well as to the accumulation and use of operational data and know-how. This may seem obvious, but it struck me that the gap between “what is possible” and “what is economically viable” may be precisely where the technical challenges and business opportunities for physical AI lie.
I asked generative AI to examine Anthropic’s research report “What work can robots do?” and related information, and to explore in depth how Japan can gain a competitive edge in this field. Please refer to the resulting analysis. When reading it, please bear in mind that the research and analysis generated by AI are based solely on publicly available information, do not necessarily reflect the actual situation, and may contain errors.

0 Comments

よろず知財コンサルティングのブログ