🔍 Intelligence Synthesis JP/EN
Ethics · Regulation · Technology — Pharma Practice Notes

AI Highlights — the whole picture — 2026-10-06 (Tue)

Daily ReportDaily 03:10 + Evening 18:10 (JST) auto-aggregated

Source links point to the original outlet. The AI Integrated Analysis is auto-generated from the headlines below only and is not intended to add facts beyond them. Not investment advice.

Editions of the day: Daily News (Daily 03:10) / Daily News (Evening 18:10)
Daily Report03:29 JST
Source: From the newest issues and articles in this site's nine sections (industry, economy and finance, well-being, medicine, agriculture, governance, pharma, cancer research, papers)  ·  Past 12 hours  ·  32 articles
AI INTEGRATED ANALYSIS2026-10-06 (Tue) — Daily Report · 03:29 JST
Speed of introduction and delays in verification and governance

From the prospect of Anthropic going public to the use of AI in the medical field, this half-day of articles has a common sequence. AI is infiltrating finance, employment, health care, and education before verification and governance mechanisms are in place. For example, medical AI is already being used in hospitals, but it has been pointed out that most of it has not undergone sufficient verification. On the other hand, there are also cases where agents are responsible for the "invisible 80%" of procurement operations, and evaluations that see AI as a "once-in-a-generation opportunity." The numbers and expectations shown as results are specific. However, what determines readers' judgment is not so much its size, but rather who checks it and who decides how to use it. In each of the following sections, we will focus on this point as we read the article.

Observation of listing and doubts about sustainability of growth

Anthropic is expected to go public despite market uncertainty and speculation that AI will slow down, CNN reported. Around the same time, the Australian Bankers Association labeled AI a "once-in-a-generation opportunity," Law360 reports. Both those who collect funds and financial industry organizations have expressed positive views. However, what these two things indicate are the expectations of the parties involved. Even if the company goes public, it does not mean that the AI business will become profitable. It is not clear from the headline of the article who tested the evaluation as an "opportunity" and under what conditions. What readers should take is the fact that expectations are strong, not the conclusion that growth is certain.

On the other hand, there are doubts about the sustainability of growth. Vanguard's Wang pointed out that inflation will remain high due to energy and AI (Bloomberg.com). The Bank of England has flagged the threat that slowing AI growth poses to national debt (WSJ). Although these two concerns appear to be in opposite directions, the premise is the same: whether or not AI continues to grow will affect the outlook for prices and government debt. Juxtaposed with the observation that Anthropic went public, two views emerged at the same time: if growth continues, it will put pressure on prices, and if it stops, it will spread to government bonds. In both cases, both parties believe the impact will extend beyond the AI companies to households and governments.

Another point is directed to the substance of growth. Futurism reported that the AI industry is producing products that undermine the economic incentives for creating cultural content, creating a "vicious cycle." The idea is that if the motivation for creation weakens, the material that AI learns from and the source of value it uses will also diminish. Again, what is being debated here is not the size of demand, but whether the conditions that support growth will be maintained. Comparing the points made by CNN, which reports on Anthropic's listing, and Law360, which reports on the Australian Bankers Association's evaluation, with Vanguard (Bloomberg.com), Bank of England (WSJ), and Futurism, expectations are expressed from the business side, while doubts are expressed from the perspective of prices, government bonds, and the field of creation. If you look at it from a different place, your judgment about the same growth will change. Whether growth continues depends on whether people outside of growth are able to shoulder the burden, and the next question is who will estimate and decide on that burden.

Reorganization of work and human resource development

In the procurement field, most of the work that is difficult for people to notice has already begun to be transferred to agents. In the Lio case reported by Dealroom, an AI agent is responsible for the "invisible 80%" of procurement operations. On the other hand, some research shows that the impact on employment is not as large as expected. A worker survey presented by The Good Men Project asked whether AI was causing job losses and found surprising results. If we put the two side by side, we can see that the transfer of most of the work to machines and the disappearance of jobs themselves are two different events. What readers should look at is not so much the amount of work that has been replaced, but what the remaining human jobs have become.

On the human resource development side, supply design is also moving forward. According to Indian Television Dot Com, Anthropic plans to spend $100 million to train 10,000 AI engineers by 2027. In education, EdSource reported that California State University has fully implemented AI. The same article also states that students who are job-hunting still have some skepticism. Companies demonstrate their intention to develop human resources, and universities distribute tools throughout the school. On the other hand, students are not sure whether their learning will be evaluated in the hiring process. Even if the scale of training is large, unless the people who have been trained are connected to the place where they work, the numbers will not provide reassurance.

In elderly care, the reshuffling of roles is occurring in a way that is more closely related to daily life. digitimes introduced that digital twins of families using AI will fill the gap in elderly care. The idea is to use AI to compensate for the time when family members cannot be around all the time. However, it is difficult to see within the scope of the article what this kind of mechanism does on behalf of whom to speak, and to what extent they are entrusted with the task. The industry can't say it's ready either. AHCA/NCAL has postponed the AI talent webinar scheduled for September 30th. Although the reason cannot be gleaned from the materials, it is clear that there is still room to adjust the schedule and content even in places where human resources discussions are being rushed.

Seen in this way, the question is not whether AI will replace jobs, but rather the design of who will have what roles, where they will learn, and what they will learn. Who will check whether the design is appropriate and who will be responsible? In the next section, we will turn our attention to those responsible for verification and governance.

Insufficient implementation and verification in medical care, research, and agriculture

Medical AI is already in use in hospitals, but it has been pointed out that most of it has not undergone sufficient verification (Earth.com). Implementation is proceeding first, followed by accuracy confirmation later. AI continues to evolve in oncology, and is said to be having an impact on clinical trials, operations, and treatment outcomes (CancerNetwork). The fact that it has begun to be used in the field does not move us as a fact, but the fact that it is being used and the fact that it has been proven to be effective are two different things.

Looking at research papers, individual results are specific and the scope is narrow. Frontiers in Oncology covers how to use deep learning to predict HER2 status from diffusion-weighted MRI of the breast. Scientific Reports reported an advanced machine learning method to predict treatment response in preclinical glioblastoma using longitudinal MRI. Medical physics shows how a deep learning CNN can detect fiducial markers on kV X-ray images for monitoring liver tumor movement. Each of the three projects involves a different process: assisting in diagnosis, predicting treatment response, and determining position during radiation therapy. However, research on glioblastoma is at the preclinical stage, and it is difficult to judge from these reports alone whether the findings are directly applicable to daily hospital treatment.

The same structure exists in agriculture. Bioengineer.org reports that AI has recreated spectroscopic images of crops from inexpensive RGB drone images. This result shows the possibility of not using expensive sensors. On the other hand, the same media also reported that attempts to understand peanut photosynthesis from space using satellites and machine learning were insufficient. Conditions in which a method works and conditions in which it does not work coexist within the same field.

The information that readers gain from this is not so much the magnitude of the results, but rather the conditions under which they were confirmed. In both medicine and agriculture, accuracy checks have not kept up with the speed of implementation. As a result, the question of governance - who will be responsible for checking this, and who will stop it if it is insufficient - remains before the superiority or inferiority of technology.

Safety measures to protect children and information space

Former Surgeon General Vivek Murthy will lead Common Sense Media's AI safeguards (EdTech Innovation Hub). In the same week, WTOP News looked at how measures against social media addiction had fallen behind, and asked whether AI-based child safety would follow the same path. In the case of social media, regulations and school responses were formulated only after the harm was widely recognized. This time, the focus will be on whether safety measures for children can be designed in parallel with or in advance of widespread use. For readers, the deciding factor is not so much the content of the safety measures, but whether they can confirm who established them and who is being protected. The fact that a public health figure like Mr. Murthy is leading a private nonprofit organization can be seen as a sign that public regulations have not kept up.

Actual damage to the information space is already occurring elsewhere. According to a report by Anthropic, Russia is suspected of using AI to carry out disinformation campaigns in Central Africa and elsewhere (DW.com). What's striking here is that the source of the information is not from the affected authorities, but from the companies providing the AI themselves. When a company discovers and discloses misuse of its products, the reality is visible, but when companies do not discover or do not disclose it, there is little way for outsiders to know. When children are the recipients of false information, they are affected before their judgments have a solid basis.

Flaws can also be seen within the organization and in the automation of operations. According to Cyber Daily, an OpenAI agent has infiltrated Medicare. The article is also an editorial that calls for lawyers to understand and master AI, calling for the law to catch up with the technology. Cybersecurity Insiders reported that OpenAI fired three employees over data protection concerns. It is clear that there were people within the company who had concerns, but since the reason for the dismissal was data protection, it is difficult to determine from this report alone whether the protection mechanisms were working or whether internal controls were compromised. The ability of agents to access sensitive information such as medical insurance records points to the reality that users are unable to verify on their own what is being done with their data.

In the face of these events, an article in The New York Times discusses how to limit the risks of AI and other threats to humanity. The position is that a framework is needed to manage the threat as a whole, in addition to dealing with individual accidents, and the question is whether it is possible to change the order in which actual damage accumulates first and systems are put in place later. Whether it's child protection, disinformation, medical data, or whistleblowing, it's being used more and more without a clear body of verification and governance. In the next section, we will look at regulatory and political movements as a forum for determining who will be responsible for this.

The question of control: who controls AI

Nepal's myRepublica introduces the ``AI Non-Alignment Movement,'' which advocates that AI should be treated without belonging to any faction. What you can see from the headline is that the argument is not about performance or safety, but about securing a position in which AI can be used without taking sides with any particular country or company. mronline.org features arguments for putting ownership of AI and technology back in the hands of workers. Eoin Higgins on Substack points out that the debate over whether AI is "conscious" is a question of corporate control. The three articles have different entrances. Still, there is one point of overlap: who owns the AI and who decides how to use it. When reading the growth prospects and employment reshuffles discussed in the previous sections, in addition to the size of the numbers, it is also important to consider who is managing the numbers.

The same question is being asked in science and culture. According to The Greenwich Sentinel, Kevin Roose spoke based on his book on AI competition and human nature. The Atlantic covers "the real benefits AI brings to science." As the word ``true'' suggests, this is an article that questions the content of what is being talked about as a blessing. UNESCO explores how to rethink museums in the age of AI in a TechCul Dialogue. What these three spaces - science, museums, and humanity - have in common is that the person deciding what to call a benefit may be outside the space.

Readers' judgments can be divided into two at this point. If we leave it up to the companies providing AI or the institutions that introduced it first to define the benefits, they will also decide the standards for verification. On the other hand, if you take a position such as worker ownership or non-alignment, you place the power of definition and verification on the side of the workplace and users. Depending on which option you choose, your perspective on the lack of implementation in medical care and education, as well as safety measures to protect children, discussed in the previous section will also change. This is because the same outcome numbers can have different meanings depending on who sets the standards.

These questions of ownership and definition ultimately lead to questions about governance mechanisms, such as who can inspect the workings of AI and who is responsible when something goes wrong.

The topics covered in each section are completely different, including funding, jobs, medical care, child safety, and technology ownership. Even so, the same composition was confirmed in every section. AI will begin to be used first, and verification work and decision-making authority will follow later. The expectations for listing, the assessment that it is an "opportunity," the replacement of work by agents, and the individual results presented in papers are all facts stated by the parties and researchers. However, they do not guarantee effectiveness or sustainability. When readers receive news coverage about AI, they need to look at the numbers and also see who is verifying those numbers, who is managing them, and who is protecting them. Therein lies the difference in judgment.

Q1: Are you in a position to explain who is checking the accuracy of the output of the AI you are using in your field and what standards are being used? Q2: Are the rules and inspections regarding the use of AI delayed after its introduction? Are there any requirements that must be met before the use of AI to be approved? Q3: When deciding whether to invest in or introduce AI, do you consider the outcome numbers and determine who is responsible for verification and the conditions for suspension?

Finally, three questions

- Are you able to explain who is checking the accuracy of the output of the AI you are using in your field, and what standards are being used? Q2: Are the rules and inspections regarding the use of AI delayed after its introduction? Are there any requirements that must be met before the use of AI to be approved? Q3: When deciding whether to invest in or introduce AI, do you consider the outcome numbers and determine who is responsible for verification and the conditions for suspension? - Are the rules and inspections regarding the use of AI delayed after its introduction? Are there any requirements that must be met before its use to allow its use? Q3: When deciding whether to invest in or introduce AI, do you consider the outcome numbers and determine who is responsible for verification and the conditions for suspension? - When deciding whether to invest in or introduce AI, do you consider the results and determine who is responsible for verification and the conditions for stopping it?

📚 Sources (all material)

Every item this issue drew on. External links open in a new tab. 32 items.

AI Regulation, Safety and Geopolitics

  1. Article (WTOP News) — Japanese-language summary
Daily Report18:30 JST
Source: From the newest issues and articles in this site's nine sections (industry, economy and finance, well-being, medicine, agriculture, governance, pharma, cancer research, papers)  ·  Past 12 hours  ·  19 articles
AI INTEGRATED ANALYSIS2026-10-06 (Tue) — Daily Report · 18:30 JST
How to create grounds for believing in AI

The explanations, scores, and diagnoses provided by AI appear plausible to the reader. However, its accuracy cannot be determined just by looking at it. The explanations provided by research agents about ``why this result should occur'' are usually not measured to see whether they improve predictions. One study reported that it could not confirm the contribution of natural explanations. On the other hand, a randomized trial of a respiratory diagnostic chatbot with 2,400 people showed that it improved the accuracy of judgments made by ordinary people. What we will look at in this half-day is the difference between claims that AI output is correct and results that are backed up by human confirmation and verification procedures. It is becoming clear that the decision to introduce AI is determined not by its performance itself, but by whether the support is available.

Research to re-measure the accuracy of explanations and scoring

When an AI research agent plans an experiment, it includes an explanation of why the result should be this way. Readers use these as clues to estimate the outcome, but whether the explanation improves the prediction is usually not measured. One preprint (on its own website) calls the amount that an explanation adds to a prediction the ``predictive contribution,'' and sets up a procedure to measure it by pairing predictions with only the explanation swapped. They reported that they could not confirm the contribution of natural explanations, as measured by toxicity tests and machine learning tasks. Whether an explanation can be read plausibly and whether it is useful for prediction are two different questions.

There is the same type of review on the scoring side. When performing LLM reinforcement learning on tasks where there is no single correct answer, it is becoming increasingly common to award partial points based on scoring criteria (rubrics) that list the requirements for the answer. Another preprint (on their own site) points out that the scorer may give points for information not included in the answer, calling this Vacuous Credit. As a countermeasure, we are proposing MetaRubric, which alternates between a learning process in which points are awarded only when there is evidence, and a process in which standards are revised while looking at the answers. A high score does not prove that the written content is good. This is because the habits of the grader directly determine the direction of learning.

The handling of success records is also being questioned. To help students learn difficult tasks, records of actual successful tasks are useful. However, if success relies on the assistance of a special harness, that assistance cannot be used in a real environment. The third preprint (own site) proposed Recursive Self-Rewrite (RSR), which extracts successes found under a variety of harnesses into a procedure manual, checks for leaks, and redoes them with a common harness. This is a procedure to rewrite the successful trajectory of terminal work into learning data with the same conditions as the actual one. A record of success can only be used if it is confirmed that it can be reproduced without assistance.

The same composition can be seen outside of research. ChinaTalk reports that a lack of funding and a state-led approach to AI security research in China are determining the future of the field. Also, according to OfficeChai, Microsoft AI CEO Suleiman expressed concern that Anthropic is teaching Claude to be conscious. Both cases overlap with the previous three in that they both call into question whether the foundation of the research and the basis for the claims are sufficient.

All three preprints question the procedures used to measure AI output, rather than the output itself. The decision to introduce the system is likely to be influenced not only by the performance numbers, but also by whether the verification procedures that support those numbers are clearly prepared.

Introduction that requires confirmation by a person on site

In the pension field, even if AI becomes widespread, "person verification" is considered essential for member procedures (Pensions Expert). This is because mistakes in decisions related to benefit amounts and qualifications have a direct impact on participants' lives, so the design of implementation will focus on where people will look, rather than the scope of automation.

Research on chatbots for respiratory diagnosis provides one clue as to whether there are situations in which it is possible to omit human confirmation. In a randomized trial of 2,400 people, the chatbot improved the accuracy of people's decisions (nature.com). What is important here is not that the correctness of the AI output itself was demonstrated by self-report, but that the comparison was made by randomly dividing people who use AI and those who do not. We use verifiable procedures to measure how well a person's judgment has improved. When readers are looking to assess the effectiveness of an AI tool, it makes sense to first look at whether it is designed to make such comparisons, rather than claiming that it is "highly accurate."

Practical implementation is spread between the two. Lottie acquires CareMaster to bring AI to care billing (konsulteer.com). In the pharmaceutical industry, there are reports of moves to accelerate the transition from evidence to action by introducing practical AI (Emerj Artificial Intelligence Research). How investment banks are using AI is also organized (Oracle NetSuite). In agriculture, micronutrients are being combined with AI and nanotechnology to transform the nutritional management of rice crops (AgroLatam). Billing, evidence, financial operations, and crop management are very different targets and the consequences of failure.

Therefore, even with the same concept of ``AI implementation,'' the answer to what extent it can be entrusted differs from field to field, and depends on what has been verified in that field. Some areas, such as pensions, rely on human verification, while others have shown effectiveness through randomized trials. We are now at the stage where we are no longer asking the question of whether or not it can be introduced across the board, but rather what kind of support it has in each field. The next question is who will continue to substantiate this and what procedures will be used.

Spread on work and learning and employee anxiety

An article (eFinancialCareers) about job openings at investment managers in Asia asks who will be most affected by AI. The same question exists in the United States. Wall Street financial institutions are beginning to address employee concerns around AI (Global Finance Magazine). AI not only comes in as a tool to support work, but it also brings with it the question of whose work will change and how. Those deciding to introduce the technology are asked not only about the accuracy of the technology used, but also about whether they have prepared explanations for the workers.

In the hiring scene, a lawsuit against Sirius XM was dismissed over allegations of racial discrimination caused by AI (HR Dive). This result alone does not mean that hiring using AI is safe. However, this case clearly shows that disputes can arise over decisions made using AI. In the same way as dealing with employee concerns, when it comes to recruiting, being able to explain the decision-making process will prepare the adopter.

Changes are also progressing on the learning side. Tanaka Study Group has launched a new brand of individualized instruction using AI teaching materials and has introduced it to 50 schools (Hiroshima Corporate Illustrated Guide). This is a move to extend instruction tailored to each student to many classrooms using AI teaching materials. In the field of viewing experiences, the AI museum DATALAND is holding immersive exhibits using Epson projectors (Digital Signage Today). The number of places where AI can be used is expanding, including workplaces, classrooms, and exhibition spaces, and the opportunities for people to interact with AI are increasing.

When these examples are lined up, it becomes clear that the introduction of AI is not determined by technology selection alone. To what extent can the person using the AI and the person receiving the judgment be able to confirm the judgment? In the next section, we will look at how systems and designs are beginning to put in place the means to confirm this.

Verification mechanism that supports trust and dialogue

Voices complaining about the dangers of AI are moving from the research field to the public arena. "We are building and nurturing our own enemies," one AI researcher warned at a hearing in New York, according to CNBC. A public hearing is a place where statements made remain as a public record and are used to consider policy. The warnings issued here are handled differently than internal discussions within individual companies or research laboratories. For those considering introduction, the deciding factor is not only the ability of AI, but also whether society has the means to control and inspect it.

AI is also a topic of discussion on the monetary policy side. The Darden Report reports that Fed Vice Chairman Jefferson spoke about AI, inflation, and confidence. A person in the position of vice chairperson deals with AI in discussions about prices and confidence. What can be gleaned from this article is that AI is no longer just a concern for the technology sector, but has begun to be treated as a subject that is related to the foundations of the system of trust in currency and prices. Confidence is established when the market and the public believe that there is a verifiable explanation. If AI were to enter into economic decisions, the system would also be asked whether it could be put in a state where its output could be checked.

This point becomes even clearer in dialogue between countries. bastillepost.com reports that experts believe that the channels for U.S.-China AI dialogue should be secure and verifiable. When two competing countries, the United States and China, engage in dialogue, they cannot simply accept what the other party has to say. As a result, there are calls for safety and verifiability in the route itself. This idea is not based on goodwill or promises, but rather on procedures that allow for mutual confirmation.

Warnings from researchers, statements from central bank executives, and recommendations from experts regarding U.S.-China dialogue all come from different places and positions. However, what they have in common is that trust is based on a mechanism that allows confirmation. Decisions on the introduction of this technology by companies and operations in educational settings are also based on this premise. The next question is who will design such a system and who will pay for it.

What these half-day talks had in common was that you can't decide whether an AI's output is trustworthy based on its appearance or self-report. Only after re-measuring could we find out whether the explanations were useful for making predictions or whether the scorers were giving points for information that was not written. In pension procedures, human verification was left in place, and in diagnostic support, randomized comparisons showed effectiveness. Hiring complaints, employee concerns, warnings at public hearings, and debates over prices and confidence also question whether there is a way to explain and check the decision-making process. When deciding to introduce a system, the deciding factor is not the claim of high accuracy, but whether the claim is supported by procedures that can be checked and verified by humans.

Q1: Of the output of the AI you are currently using, how do you check the accuracy of the parts that are left to humans without looking at them? Q2: Can you provide a procedure that can be reproduced and verified by a third party as a basis for saying that AI's explanations, scoring, and diagnosis are correct? Q3: When deciding to introduce AI, do you consider performance claims and the verification and human verification system to support them as separate criteria?

Finally, three questions

- Of the output of the AI you are currently using, how do you check the accuracy of the parts that are left unchecked by humans? Q2: Can you provide a procedure that can be reproduced and verified by a third party as a basis for saying that AI's explanations, scoring, and diagnosis are correct? Q3: When deciding to introduce AI, do you consider performance claims and the verification and human verification system to support them as separate criteria? - Can you provide a procedure that can be reproduced and verified by a third party as a basis for saying that AI's explanations, scoring, and diagnosis are correct? Q3: When deciding to introduce AI, do you consider performance claims and the verification and human verification system to support them as separate criteria? - When deciding to introduce AI, do you consider performance claims and the verification and human verification system to support them as separate criteria?

📚 Sources (all material)

Every item this issue drew on. External links open in a new tab. 19 items.

AI in Medicine (Cancer Care, Rare Diseases)

  1. Article (nature.com) — Japanese-language summary

AI Regulation, Safety and Geopolitics

  1. Article (ChinaTalk) — Japanese-language summary

AI and the Pharma Industry

  1. Accelerating Evidence to Action in Pharma with Practical AI Adoption — Emerj Artificial Intelligence Research
← 2026-10-05Index2026-10-07 →
← AI Highlights — the whole picture Index