AI Highlights — the whole picture — 2026-10-21 (Wed) Daily News

The effects of AI implementation will be first reported to the company as a result report. NAB announced 2,500 job cuts due to "rapid technological change" and in pharma, collaboration between Insilico Medicine and Servier reached its first milestone. However, it is unclear from the published materials the portion of the reduction that AI will be responsible for, nor the amount paid. What the report shows are the results of those who introduced the system, and the preparation of those who receive it remains a separate issue. National regulations covering employment impacts, labor arrangements, supervisory checks, and sources of supply. These are not being put in place as quickly as they are being introduced. The following sections should be read from the perspective of where and by whom the differences between reporting effects and perceived systems are verified.

Personnel reduction and pushback from labor and management

NAB announced it would cut 2,500 jobs due to "rapid technological change" (Information Age | ACS). 매일경제 reports that in addition to the gaming industry, where more than 10,000 jobs have been cut this year alone, there are growing concerns in communications and media that the proliferation of AI agents will affect the jobs of human employees. NAB's announcement only refers to "technological change," and the part of the reduction that AI will play is not clear from the materials. Still, the effects of introduction are explained in terms of the number of employees in several industries, including banking, gaming, communications, and media. When reading a company's results report, it is necessary to look not only at the efficiency figures, but also at how many people were hired behind the scenes.

The workers are not just accepting the situation silently. In Ontario, Canada, public sector unions have called on the province to protect AI in the workplace (Global News). According to 매일경제, there are also an increasing number of cases in which clauses regarding the introduction of AI are included in labor-management negotiations (매일경제). It is likely that the announcement of reductions will be reported first, and negotiations to determine the conditions for introduction will begin later. Even the same technology can have different effects on the workplace depending on whether a system is in place for labor and management to decide when and to what extent AI should be introduced.

The impact is uneven. Bridge Michigan treats AI entering nursing care settings as a subject whose evaluation is divided depending on whether it is viewed as ``help'' or ``danger'' (Bridge Michigan). Meanwhile, in New Mexico, an AI materials startup by Los Alamos scientists was reported to bring 80 jobs to Nambe Pueblo (Albuquerque Journal). The company will cut 2,500 or more than 10,000 people and hire 80 new people during the same period. The introduction of AI is occurring in some places where jobs are being reduced, some where jobs are increasing, and where the evaluations are divisive.

The question that will be addressed in the next section is whether this difference is determined solely by the performance of the technology or whether it also depends on the arrangements and confirmation mechanisms of the receiving side.

Progressing alliances in pharmaceuticals and disparities in infrastructure

Collaborations centering on AI continue in pharmaceutical R&D and clinical trials. Insilico Medicine and Servier have achieved their first milestone in joint research and development using AI, resulting in a milestone payment. However, the amount has not been disclosed (TradingView). On the clinical trial side, Oracle will partner with AstraZeneca to deploy AI. Expected to speed up patient screening (BigGo Finance). Vistagen selects GATC Health to apply predictive AI to flagship clinical-stage drug discovery program (BioSpace).

The three cases involve different stages of using AI. The Insilico and Servier case is different from a case that ends with just the announcement of a contract, in that the payment was made after the joint research achieved a milestone result. However, since the amount is not disclosed, the magnitude of the results cannot be measured from the outside. The two cases, Oracle and AstraZeneca, and Vistagen and GATC Health, are announcements of introduction or selection, and the effects are still being confirmed. Rather than counting the number of announcements, it is more useful to discern what has been confirmed at what stage when making investment and partnership decisions.

Even if you can choose a partner to partner with, it doesn't necessarily mean that the recipient is ready. Manufacturing Chemist reports that there is a gap in the digital maturity of the pharmaceutical industry, which is hindering the adoption of AI. This means that even if the same AI is used, the results obtained may differ between companies with advanced data and operations and those without. This material does not include numerical values that measure this difference. Therefore, when reading an announcement of a partnership, it is necessary to separately check how well established the partner company's infrastructure is.

The results of AI will depend less on which technology is chosen and more on the preparation of those who receive it. This development goes beyond data and business infrastructure. As the introduction progresses, the question becomes whether systems such as supervision and safety confirmation are being put in place at the same speed.

Supervision regarding finance/macro and fairness

The Bank of England is considering the impact of "stress testing AI" on the financial sector, reports Claims Journal. Stress testing is a verification procedure carried out by authorities to confirm whether financial institutions can withstand sudden changes in the environment. The addition of AI to the list of considerations indicates that not only companies implementing AI but also those supervising it need to confirm their work. The same interest exists in the technology field. OrcaRouter's "AI Red Team Explained" explains what a red team actually tests. Red teaming is an evaluation role that proactively searches for weaknesses in a system from the perspective of an attacker or abuser. For those deciding to introduce the product, there is a difference between being able to show that the product is ``said to be safe'' and ``what was tested under what conditions.'' When receiving explanations from vendors, this is the basis for requesting disclosure of the scope of testing and results as a condition.

kmop.gr's ``AI, Data, Justice'' asks how we can encourage equality and fairer decision-making. What is dealt with here is the situation where AI is used to allocate people and projects, and to support decisions. If the data on which a decision is based is biased, the results may also be biased. Therefore, fairness cannot be confirmed just by looking at the output; it is necessary to trace back to what data is used and who can explain how. The questions in this article overlap with the direction of demanding accountability from organizations that use AI. Stress tests check the resilience of financial institutions, red teams check the weaknesses in technology, and fairness discussions check the basis of judgments.Although the targets of the three tests are different, they all agree that the users are responsible for the verification.

Putting this need for oversight into a larger context is Vocal's "Three decisive crosscurrents: AI, energy, debt, and the global economy at a tipping point." As the title suggests, the article treats AI not as a standalone technological innovation, but as a macroeconomic factor alongside energy and debt, suggesting that the global economy is at a tipping point. However, what can be confirmed here is only the title and the author's perspective, and it cannot be determined whether or not the change will actually occur. Still, if AI works in conjunction with funding, electricity demand, and financial conditions, individual companies' decisions to introduce AI will become inseparable from financial stability and social fairness. This shift in the order in which companies report their results first, and mechanisms for verification and explanation are put in place later, could pose a problem not only for individual workplaces, but also for the stability of the economy as a whole.

What the materials presented so far have in common is that the measures for measuring the effects of AI and the development of systems to capture those effects are progressing at different speeds. The difference between success and failure in implementation is whether a system for verification and explanation was prepared at the same time as reporting results.

US-China conflict and questions of values

China has condemned Washington's restrictions on AI companies and warned of retaliation, BusinessToday Malaysia reports. Three moves have been reported: restrictions from the United States, condemnation from China, and warnings of retaliation. What we can see here is that restrictions on AI companies are not just a matter of competition between companies, but are also a source of retribution between countries. From the headline of the article, it is not possible to confirm the names of the companies subject to restrictions or the details of the specific measures taken. Therefore, it is not possible to say for sure which companies will be affected and to what extent.

Still, it is of no small significance to those considering its introduction. Until now, when deciding to procure or partner with AI, the main comparisons have been performance, price, and internal operational structure. Added to this is the question of which country's regulations and countermeasures the supplier is affected by. Given that China has reportedly warned of retaliation, it is necessary to consider the possibility that the range of available technologies and business partners may change if the exchange of restrictions and countermeasures continues. However, the report does not indicate what form the retaliation will actually take. It would be appropriate to approach the decision while keeping the confirmed facts and unknown points separate, without filling in the blanks with speculation.

On the other hand, there is also a movement to question the use of technology from the perspective of value. The Catholic Times featured “AI and Evangelization.” What the religious community is debating is not whether AI can be used, but how to position AI in the work of transmitting faith. Here, rather than efficiency or processing power, the question is what to use it for and for whom. If conflicts between nations are about who owns technology, then the debate over evangelization can be said to be about what technology should be used for. This is all that can be gleaned from the headline of this article; it is not possible to read the official position of the church or the presence or absence of specific guidelines.

Although the two stories take place in different places, they have something in common. In both cases, the point of contention is not the performance of the technology itself, but the power to enclose it and the standards that determine how to use it. Enclosure depends on the circumstances of the country, and the standards for use must be determined by organizations and communities themselves. The same is true for corporate implementation; before selecting a technology, the organization needs to decide how to estimate supply uncertainty and what value to protect. The presence or absence of this preparation will determine the success or failure of the next introduction.

The events covered in each section vary by industry and region. These include layoffs and pushback between labor and management, disparities in pharmaceutical partnerships and infrastructure, financial authorities' review and safety testing, and U.S.-China restrictions and warnings of retaliation. What they have in common is the order in which the effects of the introduction are discussed first, and the mechanisms to absorb them follow later. Companies can produce numerical results, but explanations of the impact on employment, the form of agreements, disclosure of the scope of testing, and understanding of supplier circumstances cannot be verified unless organizations and systems are in place. What determines the success or failure of an introduction is not so much the performance of the technology itself, but the extent to which preparations have been made by the person implementing it. When reading an achievement report, it is necessary to look at whether it shows the system that supports it, along with the numbers.

Q1: When AI is introduced into your work, have you agreed with whom and in what form before introducing it what will change and what will not change? Q2: When introducing AI is described as "safe" or "fair," do you have a procedure for disclosing what conditions and results were tested? Q3: When talking about the results of AI implementation, are you able to show in the same report the employment that took place behind the scenes and the impact of regulations on suppliers?

Finally, three questions

- When AI is introduced into your work, have you agreed with whom and in what form before introducing it what will and will not change? Q2: When introducing AI is described as "safe" or "fair," do you have a procedure for disclosing what conditions and results were tested? Q3: When talking about the results of AI implementation, are you able to show in the same report the employment that took place behind the scenes and the impact of regulations on suppliers? - When introducing AI is described as ``safe'' or ``fair,'' do they have a procedure for disclosing what conditions and results were tested? Q3: When talking about the results of AI implementation, are you able to show in the same report the employment that took place behind the scenes and the impact of regulations on suppliers? - When talking about the results of AI implementation, are you able to show in the same report the employment that took place behind the scenes and the impact of regulations on suppliers?