The Digital Deficit and the 'Year of the AI Agent' — Three Structural Biases Settling in the Draft Phase of Japan's Digital Society Vision Council

Japan's Digital Society Vision Council has entered the draft phase of the next Priority Plan for the Realization of a Digital Society. The 11th meeting on December 4, 2025 and the 12th meeting on May 22, 2026 are the two sessions where this draft is being worked among members. This column reads, from ISVD's perspective, the contours of debate that are settling in this phase.

At the 11th meeting (December 4, 2025), under Chairman Jun Murai, members from local government, industry, research, and civil society participated alongside Minister Matsumoto for Digital Transformation, Parliamentary Vice-Minister Kawasaki, and State Minister Imaeda. Main topics included digital talent development, domestic AI development, response to the digital deficit, data governance, and citizens' digital literacy (with the 45% online completion rate of the national census raised as a reference point).

At the 12th meeting (May 22, 2026), materials were submitted by Mayor Ikeda of Miyakonojo City (Miyazaki), Mr. Mikitani of Rakuten Group, and Governor Muraoka of Yamaguchi Prefecture. Mr. Mikitani centered his submission on five points: strengthening AI competitiveness, domestic AI, periodic review of government-adopted models, and securing both domestic and foreign talent. Miyakonojo and Yamaguchi brought field perspectives from local-government implementation.

A shared context across both meetings is that the digital deficit is referenced as a starting point of debate. The 2024 figure was ¥6.6507 trillion (international balance of payments statistics); some analyses project expansion to roughly ¥45 trillion by 2035. Corporate payments for generative AI services, cloud subscriptions, and advertising technology are discussed as the main components.

Of the materials distributed in the meetings, only three submissions are publicly released. The Priority Plan draft (summary and draft text) and the FY2025 awareness survey are handled as member-only materials. The final Priority Plan is fixed by Cabinet decision after public comment.

The "digital deficit" indicator extracts digital-related items from the services balance: cloud computing, digital advertising, software licenses, and data transmission are the main constituents. The indicator itself is relatively new — the Ministry of Finance and the Bank of Japan began compiling it from the international balance of payments in the early 2020s.

The deployment of this indicator has expanded in response to two policy needs. One is the desire for a KPI that frames the priority of digital industrial policy in monetary terms, which is easy to discuss. The other is that it makes "high foreign dependency" visible, supporting arguments for investment in domestic digital infrastructure.

The discourse of "the Year of the AI Agent" refers to the situation in which generative AI is expanding from one-off text and image generation to autonomous task execution. The phrase took hold in industry in 2025, and response to the AI-agent era became a central theme at the 11th and 12th meetings. Mr. Mikitani's submission rests on this context.

ISVD, on the other hand, has worked through its cognitive debt series on a different scope: the debt that the social penetration of AI imposes on long-term human learning, labor, and judgment. EEG experiments at MIT Media Lab, deskilling among colonoscopy physicians, and AI hallucinations in legal filings illustrate how AI dependence erodes cognitive capacity. None of this is addressed within a competitiveness frame.

When we look at the contours of debate solidifying in the draft phase across the 11th and 12th meetings, three structural biases come into focus.

Bias 1: The one-directional nature of the "deficit" indicator. The digital deficit measures balance differences. It is easy to discuss in monetary units but does not capture transfers of knowledge or productivity, nor the social benefits generated by digital consumption. While payments to foreign clouds are visualized as "deficit," the operational efficiency, new businesses, and educational opportunities created domestically by using them are dispersed across different statistics and hard to see. Because the indicator has sensitivity only in the "excess payment" direction, countermeasures also tilt toward "domestic substitution" and "in-house construction." In ISVD's reading, the digital-deficit discussion needs to set, alongside the deficit figure, equivalent measures for "benefit transfer," "domestic productivity gains," and "expansion of educational opportunity." A one-directional indicator produces a one-directional policy.

Bias 2: Enclosing AI in a competitiveness frame. The debate at the 11th and 12th meetings positions AI as "decisive for Japan's international competitiveness." Mr. Mikitani's submission is the clearest example. Within this frame, the question of AI policy converges on "how fast and how widely to deploy." But the social penetration of AI has other scopes. Cognitive debt accumulating in human cognition, judgment, and learning; AI systems accelerating epistemic injustice (concentrated amplification of misinformation, biased distribution of information sources); the autonomous behavior of AI agents producing ambiguity in responsibility allocation — these cannot be measured in "amount of deployment." They require different indicators for "what is lost by deployment." A design enclosed in a competitiveness frame excludes this other scope from policy vocabulary.

Bias 3: The narrowness of decision-making channels. The draft is worked among members, and voices from local government enter only as "success stories" (Miyakonojo's My Number Card issuance rate among the highest in Japan, Yamaguchi's outbound communication on digital infrastructure). The awareness survey is handled as a member-only material; summary and interpretation are released only through the Digital Agency. By the time public comment opens, the main structure of the draft is already fixed. Municipalities that have not yet engaged (population under 5,000, no published council minutes, no digital support officer placed), the front lines that absorb the load of AI deployment (the final touch-points in education, medicine, and welfare), and the low-wage labor that supports digital consumption (content moderation, data labeling) are all difficult to make visible through this channel.

The three biases are not independent. The one-directional "deficit" indicator makes "domestic substitution" the main axis of debate, which reinforces the "AI competitiveness" frame and ultimately seats "successful municipalities and large corporations" at the center of the agenda.

10th

2025-04-22

Review of current Plan

→

11th

2025-12-04

Initial draft of next Plan

→

12th

2026-05-22

Draft review (member-only materials)

→

Public comment

(TBD)

Main structure already fixed

→

Cabinet decision

(TBD)

Fixed as policy document

Worked among members

• Digital talent
• Domestic AI
• Digital deficit
• Government Cloud
• Voices of successful municipalities

Complementary issues from private/nonprofit

• Cognitive debt
• Digital benefit transfer
• Unengaged municipalities
• Low-wage digital labor
• End-point practice

To complement the biases of the draft phase structurally, independent research and practice institutes like ISVD need to keep presenting the following issues through separate channels.

First, building digital-benefit indicators. Alongside the "deficit," private actors should construct and openly publish indicator sets measuring "benefit transfer," "domestic productivity gains," and "educational opportunity" on equal footing. Second, expanding the scope of AI policy to include cognitive debt. Issues that the competitiveness frame does not address should continue to be sent into the discussion through other channels. Third, making visible the municipalities that have not yet engaged, the front-line workers at the final touch-points, and the low-wage digital labor that powers digital consumption. Voices that fall out of the central council's channels must be picked up through field interviews and data collection.

The Priority Plan is fixed as a Cabinet-decided policy document. But the "digital society" that document portrays is shaped by what is included and excluded at the entry of debate. Critically detecting the biases of the draft phase and presenting complementary issues to society is one of the central roles of private and nonprofit practice of social vision.

• Reclaiming "Social Vision" from State Policy Vocabulary — Actor, scope, and method differences between ISVD and the council
• Digital Society Vision Council, 12th Meeting — Briefing — Issues from the May 22, 2026 meeting
• What Is Cognitive Debt — The debt accumulating against human capacity in the AI era

References

11th Meeting of the Digital Society Vision Council — Digital Agency (2025). Digital Agency Official Site

12th Meeting of the Digital Society Vision Council — Digital Agency (2026). Digital Agency Official Site

Japan's Digital-Related Balance (2025) — Large Digital Deficit Continues, Productivity Gains via AI Are Urgent — Mitsubishi Research Institute (2026). MRI Economic Insight

The Digital Deficit in the Generative AI Era — How Should Japan Respond — nippon.com Editorial (2025). nippon.com

Opinion Submitted to the 12th Meeting of the Digital Society Vision Council — Japan Association of New Economy (2026). JANE

Recommended Reading

Shigeki Uno, 『民主主義とは何か』 (What Is Democracy?), Kodansha Gendai Shinsho, 2020. A reading of decision-making channels and the question of "whose voice gets in" through both intellectual history and contemporary politics. A foundational text for reading what is happening in the draft phase of the Priority Plan within the wider context of democracy.