The Future of AI Material Review — Dividing the Work Between People and AI── The principles that must hold, no matter what changes

01Technology Trends — The AI That Will Reach Review

First, let us line up, in honest terms, the technologies likely to reach the material-review floor. Not "already usable," but "getting closer." Misjudge that distance and the future becomes mere wishful thinking. Four directions are growing right now.

What the four share is that AI is shifting from "a tool that lines up words" to "a tool you can trace back to evidence." That said, however much accuracy rises in each, the nature of predicting by probability itself does not vanish. It becomes faster and easier to trace. But the room to err quietly remains. This premise does not break down anywhere in the picture of the future.

02Dividing the Work Between People and AI — What to Delegate, What to Keep

Once technology arrives, the next question is the division of labor. What to lean toward AI, and what to keep with people. A useful way of thinking here is the phrase the American Medical Association (AMA) has used from early on: augmented intelligence (= intelligence that augments human judgment rather than replacing it). The AMA called medical AI not "artificial intelligence" but "augmented intelligence," positioning AI as a tool that helps the physician rather than replacing them. In material review, that line holds just as directly.

Set out plainly, the work that can lean toward AI and the judgment that stays with people divide as follows.

The left column is work that is high in volume, fixed in form, and where speed counts. The right column is checking against the approved scope, the read that pictures the recipient, and the taking on of responsibility — the work of deciding a single right answer in the context of the moment. However clever AI becomes in the future, this right column cannot be handed over wholesale. The reason is as we saw in Vol. 5: judging the approved scope is not a probabilistic prediction but a decision that carries responsibility. The division moves, but where responsibility sits does not.

03The Evolution of Regulation — How the Rules Move When AI Is Assumed

It is not only technology that changes. The regulatory side, too, has begun to move on the assumption that AI is entering. This part cannot be overlooked in reading the future.

Internationally, three currents stand out. First, Europe's AI Act (= Regulation (EU) 2024/1689). It classifies AI in settings such as healthcare as high-risk (= high risk) and mandates record-keeping, transparency, and human oversight. Second, the approach to managing AI/ML-based medical device software set out by the U.S. FDA. On the premise that models will be updated later, it seeks to guarantee "no getting smarter without permission" through the PCCP (= Predetermined Change Control Plan, a framework that plans and files in advance the scope and method of updates). Third, ISO/IEC 42001 (= the standard for an AI management system, 2023), the international standard for operating AI within an organization — the yardstick for record and traceability touched on in Vol. 7. In Japan, too, the Ministry of Health, Labour and Welfare (MHLW) has shown priority areas for AI use in the health and medical field and has debated the twin wheels of accelerating development and operating it properly.

Here is a point that must not be confused. Even as regulation grows finer on the premise of AI, the yardstick that measures the substance of the material itself does not change. Under the Pharmaceuticals and Medical Devices Act, the prohibition of exaggerated advertising is Article 66, the prohibition of advertising unapproved drugs and the like is Article 68, and the propriety of information provision in sales-information-provision activities is Article 68-2. This arrangement is the same whether the writer is a person or an AI. The Guidelines on Sales Information Provision Activities for Prescription Drugs (the SIP Guidelines, a notice from the Director-General of the MHLW Pharmaceutical Safety and Environmental Health Bureau, 2018), and the Standards for Fair Advertising (issued by the Director of the Compliance and Narcotics Division, Pharmaceutical Safety and Environmental Health Bureau, MHLW), do not loosen for the novelty of the tool.

04Principles to Uphold — Four Pillars That Do Not Move

Technology arrives, the division of labor shifts, and regulation grows finer on the premise of AI. Even so, there are pillars that must not be moved. The foundation of the future is not the new technology but these four pillars.

There is one more line to keep consciously, all the more as AI enters review deeply in the future. What material review protects is, above all, whether the substance of the information provided stays within the approved scope and the regulations. What a medical representative (= MR) handles is the provision of information about drugs; price, stock, delivery, ordering, and price negotiation are not that role. Those belong to the transactions and logistics between the drug wholesaler and hospital procurement — a separate system from what material review measures. It is precisely when the mood arises that "AI could be entrusted with anything around materials" that this boundary must not be blurred. In the picture of the future too, what review faces is the propriety of information, not the rationalization of transactions.

05The Patient-Centered Axis — Where Every Judgment Faces

A single axis runs through the pillars so far. It is the answer to the question of why material review exists at all. It is not to protect the organization, nor to avoid being scolded by regulators. What it ultimately faces is the safety and proper use of the patient who lies beyond the healthcare professional receiving the material. Lose sight of this axis and review shrinks to a hollow, formal inspection.

The World Health Organization (WHO) set out six principles as guidance for AI in health — protecting human autonomy; promoting safety and the public interest; ensuring transparency and explainability; fostering responsibility and accountability; ensuring inclusiveness and equity; and promoting sustainability. Line them up and none of them is about technology. They are principles that face the human being. Even if AI makes material review faster, where these six face does not move.

So however clever AI becomes in the future, the last question that remains for the reviewer does not change: can the healthcare professional who received this material make the right judgment for the patient? AI can help handle this question faster, but it cannot take on the question itself.

06Summary — The Future on a Single Page

Let us fold everything so far onto a single page. Future material review will move as follows.

• Technology arrives — multimodal, grounding in evidence, review-specific versions, agentification. AI shifts from "a tool that lines up words" to "a tool you can trace back to evidence." But the nature of erring quietly does not vanish.
• The division of labor shifts — high-volume, fixed-form work leans toward AI. Judging the approved scope, the read of validity, and taking on responsibility stay with people. AI does not replace people; it augments them.
• Regulation becomes a premise — the AI Act, PCCP, and ISO/IEC 42001 add "how to manage AI." But the yardstick of substance — exaggeration Article 66, unapproved Article 68, information provision Article 68-2 — does not change.
• The principles do not move — final responsibility is human, judgment is tied to evidence, traceable and reproducible, judge by the substance. And do not mix information provision with transactions and logistics.
• The axis does not change — everything faces patient safety and proper use. Speed and efficiency follow this axis.

When speaking of the future, the greatest danger is having your eyes stolen by the new technology and losing sight of the axis. Conversely, if you fix the axis and the principles first, then whatever new tool arrives, you can calmly decide how far to admit it. Separate what changes from what does not — this is the only sure way to prepare for the future.

07Connections to Other Chapters — How to Bind the Series Together

This installment's picture of the future extends the accumulation of every prior chapter. Finally, let us bind the whole series back into a single line. Use it as a map when you read back.

• Vol. 1 — What Is Happening Now — the map of "what changes and what does not" drawn at the starting point. This installment re-fixed that "does not change" as principle.
• Vol. 5 — Standardizing Review — a common language that reduces person-dependence is the foundation that makes the human-AI division of labor work. Even as the division shifts, with a common yardstick the judgment does not waver.
• Vol. 7 — Records, Audit Trails, and CAPA — the implementation of the pillar of being traceable and reproducible. The more AI judgment increases in the future, the more this design of the record takes effect.
• Vol. 8 — Reviewers' AI Literacy — what turns the picture of the future into reality is the capacity of people who use AI well. Upholding the principles and not losing sight of the axis both come down, in the end, to human judgment.

The era in which AI enters material review has already begun. From here, AI will become a tool that is faster and more traceable to evidence. The division of labor will shift, and regulation will grow finer on the premise of AI. Even so, what we have confirmed across these ten installments folds into one thing — the tools may change, but the principles and the axis do not move. Final responsibility is borne by a person, judgment is tied to evidence, review remains in a traceable form, and what is measured is the substance that is written. And everything faces patient safety and the proper provision of information.

There is no need to fear the future, nor to overtrust it. Admit what changes, uphold what does not. As long as you do not misjudge that sorting, AI will make material review reliably faster and reliably more careful. From the next installment, the tone changes: we move to the essay "Days of Quiet Grace" (Hibi Kore Kōjitsu), which depicts review not as a system but as a human endeavor. Now that we have finished, in one sweep, the talk of tools, it is our turn to look at the daily lives of the people who use them.