AI Highlights — the whole picture — 2026-10-29 (Thu) Daily News

At the same time, layoffs, drug discovery funding and partnerships, regulatory hearings, and a change in who people talk to are all occurring at the same time. Elementor's CEO said AI has reduced the number of employees by 60%. Anthropic has entered drug discovery, announcing six partnerships. Both are examples of progress in implementation, but what this text focuses on is not the speed of progress. How jobs, funding, and partnerships are redistributed as a result of AI inclusion, and who assumes responsibility for those outcomes. The FTC's request for explanations from executives of provider companies and doubts about the fairness of evaluations are on the same lines. Before deciding to introduce it, you should check where results and responsibilities are placed.

Personnel reduction and corporate reorganization

At Elementor in Israel, the CEO says AI has reduced the number of employees by 60% (CTech). The size of the workforce itself is an example of a numerical result of the introduction of AI. The CEO of US bank PNC expressed the view that AI will change the structure of M&A (Banking Dive). Major US IT companies are offering retirement benefits to thousands of employees, while Anthropic is reportedly preparing to go public (WKMG). Companies cutting headcount, financial institutions reconsidering their acquisition and integration structures, and AI companies heading to the capital markets are all lining up at the same time.

The movement is not limited to companies. South Korea's growth rate was 3.4%, higher than expected, due to strong demand for semiconductors (Al Jazeera). The demand for computing resources that power AI is beginning to be reflected in national growth rates. On the implementation side, Itential introduced a new platform for AI collaboration called “AI Skills” (Futuriom). As a product, we are working on the layer that connects individual AI to existing business and operational mechanisms. The increase in demand and the development of infrastructure to connect it to the field are proceeding in parallel, prerequisites for changes in personnel and organization.

Three points emerge from this that are relevant to the reader's judgment. First, the 60% reduction and retirement benefits for thousands of employees strengthen the view that the effects of AI are measured by the number of employees. However, the higher the number, the more questions will be asked about the quality of work and who will be responsible for it. Second, PNC's CEO pointed out a change in the structure of M&A, indicating the possibility that the valuation axes for buyers and buyers will change. However, how things change should be viewed as the CEO's perspective, and should not be treated as a set outcome. Third, even if demand and implementation are progressing, such as South Korea's growth rate and Itential's platform, this does not necessarily translate directly to the results of each company. Success depends on whether the implementing organization has a design that takes responsibility for operations and results.

Employment, enterprise form, demand, and implementation trends appear to be separate. However, the common question is not so much the speed of change, but rather who will bear the consequences of that change. In the next section, we will look at how this positioning of responsibility and performance is reflected in the allocation of funding and partnerships.

Funds and partnerships for drug discovery

Anthropic has made a full-scale entry into drug discovery. Fudu Niu Niu reported that it had acquired a "one-man pharmaceutical company", partnered with six major industry companies, and announced the latest R&D progress. BigGo Finance reported on Claude's publication of early drug discovery results and the acquisition as a new model for AI drug discovery. Digital Health News reports that Claude has partnered with six life science organizations to advance drug development. Depending on the media, they are referred to differently as ``six major companies'' or ``six organizations,'' but they all have one thing in common: they have six alliances.

Fund movements are also trending in the same direction. Beijing-based BioGeometry has raised over $100 million for AI drug discovery R&D (Endpoints News). This means that the movement of model development companies to start drug discovery on their own and the movement of more than $100 million to companies specializing in drug discovery have coincided at the same time. Tohoku Securities has expressed the view that the need for more efficient research and development will accelerate the industrial implementation of AI drug discovery (Moomoo). From this perspective, what is driving demand is not so much an interest in new technology as a desire to make the research and development process itself more efficient.

Reorganization of surrounding areas will also proceed. Eversana has acquired UK-based AI health economics research company Estima and combined its operations (Fierce Pharma). Estima is a company that is not involved in upstream drug discovery, but rather evaluates medical economics. What buyers needed was not just the ability to search for molecules.

What readers should look at is who these projects are connected with and to what extent. The number of alliances and the amount of procurement alone can only indicate the speed of implementation. Anthropic's acquisition and partnership announcement, BioGeometry's procurement, and Eversana's integration of Estima are all moves that determine who will receive research and development results in the form of contracts and structures. Once funds and partnerships are gathered, how are the responsibilities and results shared? We will deal with that in the next section.

Careful introduction into clinical and field settings

The point that the pharmaceutical industry is being cautious about introducing AI is cited in the article ``Cautionary attitude toward introducing AI in the pharmaceutical industry: Understanding the reasons behind the lack of progress'' in Pharmaceutical Executive. On the other hand, in the field of clinical trial management, we have begun to see the introduction of this technology in the form of partnerships. Appian and PSI CRO have entered into a strategic partnership to advance AI-powered clinical trial operations (PR Newswire). Even in a cautious industry, efforts are underway to name specific tasks such as test management. What readers should look at is not whether they are positive or negative about the introduction. The business that incorporates AI is specified, and the partner responsible for its operation is clearly identified as a partner.

Introduction after on-site verification can be confirmed in other fields as well. The FDA has approved an AI-powered chest CT software that analyzes the aorta (Diagnostic Imaging). Regulatory approval indicates that the application and performance have been verified. This is an example of how caution is not stopping people from introducing the technology, but rather choosing applications that have been verified. In agriculture, AI and drones are tracking the growth stages of hundreds of rice varieties (EurekAlert!). With a scale of several hundred varieties, it would be a burden to continue observing them manually. The role of AI here is narrowed down to one task: tracking.

In situations similar to daily life, the number that 89.3% of elderly people request voice reminders was introduced, and ways to utilize AI were demonstrated (shattered.io). What can be gleaned from this article is that the usage is specific, ``reminder,'' and the user's wishes are expressed in numbers. However, this number alone does not tell us the actual usage and effects.

When we line up the five materials, there are commonalities in the situations in which they are introduced. The usage is specifically specified, either by business, approval, tracked target, or user's wishes. On the other hand, a cautious attitude stands out in situations where the company is asked to introduce a system without knowing how it will be used and who is responsible. In other words, whether AI can be introduced or not depends not so much on what AI can do, but on what will be verified on-site and who will be responsible for the results. This perspective leads to the question of which actors are responsible for outcomes and responsibility, which we will address next.

Doubts about the eyes and fairness of the authorities

Yahoo reports that the FTC has asked Anthropic executives to testify. According to the headline, the purpose is to investigate consumer harm. Until now, companies providing AI have often been talked about in terms of performance and speed of implementation. Added to this is a movement by regulatory authorities to demand explanations from individual executives. From the perspective of those deciding to introduce the model, it is important to check before contracting or operating the model to determine what kind of investigation the provider of the model being used may be subject to. Even if you choose a product based on its speed, you may find yourself in a position where you are required to explain it later.

There are also doubts about the evaluation side. Technology Org reports that even though AI ratings may appear unbiased, they may not be fair. If being unbiased and being fair are treated as the same thing, a decision to introduce a product based on good evaluation numbers may be incorrect. Companies and local governments that receive evaluation results need to confirm what is being measured and to whom it is fair, apart from the explanation provided by the provider. Verification does not end when an evaluation is issued; confirmation work remains after that.

There are also disagreements about how the competition is structured. The Carnegie Endowment for International Peace believes that simply competing for speed will not produce results in AI, and that the United States has other options. That's all you can tell from the material headings; the details of the options are beyond the scope of this article. However, if the assumption that speed is given top priority is shaken, the judgment that the speed of introduction itself should be treated as a grade will be weakened.

The impact on elections has also been raised as a concern. The Pennsylvania State University asked experts if hackers could use AI to influence US elections. I won't go into the details of the answer because it's not in the materials, but the very fact that the question is being posed shows the need to supervise and verify how AI is used in areas such as elections, where trust in the results is at stake. The FTC's investigations, the fairness of evaluations, the nature of competition, and elections are all four areas in which the focus of decision-making is shifting from the rapid introduction of AI to who will explain when a problem occurs and who will bear the consequences. The next question is how to design this acceptance in terms of on-site operations.

Changes in people's hearts and connections

More and more people are consulting AI rather than people close to them. A survey reported by PRLeap states that ``one in three people chose AI,'' and also points out that loneliness hinders corporate innovation. This means that the way people choose who they consult with has changed. VICE reported on a new service that allows you to design your ideal "AI girlfriend." A maximum of 50 people can be created. The National Law Review predicts that the mental health AI market will grow to $8.8 billion in 2030. However, this is a prediction made by the same media and is not a confirmed figure. It can be seen that AI is moving into the position of a human conversation partner in three situations: personal consultation, relationship design, and markets similar to medical care.

There is also debate over the extent to which AI will interact with people's inner lives. The Franklin County Free Press reported that Seth and Sebo were in a dispute over AI consciousness and the welfare of the models. Whether we recognize consciousness and welfare in AI has implications for how users interact with AI and how businesses assume responsibility. In education, Bioengineer.org reported that an AI inspired by bumblebees will help students learn English by determining their emotions from their facial expressions. The ability to read emotions has entered the field of learning. In the creative field, SIDE-LINE published ``2026 Edition, 7 AI Video Tools for Electronic Music Producers''. Creators have an increasing number of tools to choose from.

Let's put the material we've learned so far into the context of a decision-making situation. In a situation where one out of three people chooses AI, creating an organization or support system based on the assumption that only people can be consulted may deviate from the reality. Looking at the service that allows you to create up to 50 AI girlfriends and the market that is predicted to reach $8.8 billion in 2030, businesses that respond to the loneliness and anxiety of users are expected to expand. At the same time, the question is who will be responsible for its design and operation. The same question arises with learning support and creative tools that read emotions from facial expressions. Even if the debate over the consciousness of AI has not been resolved, those implementing it need to decide first how much of an impact they will have on users.

As the distance between humans and AI grows closer, the question of who will be responsible for the outcome in situations involving AI becomes more important than whether or not to use AI.

The movements covered in each section reach the same point even if the field is different. Companies that have reduced their workforce have seen results shown in the numbers, but who will be in charge of operations remains a separate issue. In the area of drug discovery, funding and partnership partners have been clearly indicated, making it easier to see who will be responsible for the initiative. In the field of clinical trial management and diagnostic imaging, the system is being introduced in such a way that tasks are specified and parties responsible for management and verification are clearly specified. Questions about the fairness of interviews and evaluations by the authorities indicate that judgments cannot be made based solely on the explanations provided by the provider. Even in situations where AI is used as a counselor for people, the numbers from predictions and surveys are not final results. Rather than the speed of implementation, the deciding factor is who is designed to take on responsibility and results, and to what extent.

Q1: When incorporating AI into your work, is the person who will ultimately be responsible for the output results specified in the procedure? Q2: Apart from the provider's explanation and evaluation numbers, are there any procedures established before operation to confirm what is being measured and to whom it is fair? Q3: Are the results obtained from the introduction of AI and the resulting responsibilities assigned to the same person within the organization?

Finally, three questions

- When incorporating AI into your work, is the person ultimately responsible for the results of its output named in the process? Q2: Apart from the provider's explanation and evaluation numbers, are there any procedures established before operation to confirm what is being measured and to whom it is fair? Q3: Are the results obtained from the introduction of AI and the resulting responsibilities assigned to the same person within the organization? - Apart from the provider's explanation and evaluation numbers, are there procedures established before operation to confirm what is being measured and to whom it is fair? Q3: Are the results obtained from the introduction of AI and the resulting responsibilities assigned to the same person within the organization? - Are the results of AI implementation and the resulting responsibilities assigned to the same people within the organization?