AI Highlights — the whole picture — 2026-10-13 (Tue) Morning News

In the United States, there is talk of the possibility that AI capital investment may change from a ``support'' to a ``weight'' on the economy, and at Microsoft Israel, 50 employees in the AI agent division will undergo a pre-dismissal hearing. The former shows the burden that a reversal of investment places on the economy as a whole, while the latter shows how the costs of the reversal fall on specific sectors and occupations. The performance of AI is improving, and its introduction is progressing in concrete ways in medical care and drug discovery. However, when investment fluctuates or the number of employees decreases, who will bear the consequences and who will take responsibility cannot be determined from performance numbers. In this article, we will take a look at where the burden and responsibility is placed in each aspect of equipment, the workplace, trust, and daily life.

The weight of capital investment rebounds on the economy and region

In the US economic outlook, talk has begun to be made of the possibility that AI capital investment will shift from being a "support" to a "weight." Jevons Global's CIO points out that a reversal in AI capital spending could lead to a recession in the U.S. economy (CNBC). What I would like to note here is that this view treats the increase or decrease in investment itself as a variable in the economy, rather than the performance of individual AI companies. If the growth rate has been estimated based on the assumption that spending on data centers and semiconductors will continue to grow, the downside when that assumption breaks down will be more than just the slowdown in capital investment.

The repercussions are not limited to the United States. Concerns over the investment burden of AI semiconductors are reportedly weighing on South Korean and Taiwanese stock markets (Finimize). Both markets are deeply embedded in the supply chain of semiconductors for AI, and while they will benefit as long as U.S. companies increase capital investment, they will also move in the opposite direction if they become aware of the burden. In other words, the structure is such that US investment decisions influence asset values elsewhere, such as stock prices in South Korea and Taiwan. For readers who own AI-related stocks, what they should be looking at is less about the technological level of individual companies and more about whose profits and finances will be squeezed by the investment burden.

The importance of investment is also beginning to be questioned from the local community side. Jolie, who is running for governor of Florida, advocates for a moratorium on AI data center development (News4JAX). Data centers are facilities that require electricity and land, and will be subject to voters' decisions once capital investment figures are concreted into local development plans. Investment momentum can be constrained not only by financial markets and economic statistics, but also by electoral procedures. If plans are delayed or canceled, the outlook for capital investment mentioned above will also be affected.

On the employment front, discussions about burdens are progressing in parallel. Bill Gates has reportedly warned that AI will lead to job losses and said that North America faces competition from China and an urgent need for regulation (EnergyNow.com). Competition demands urgent investment, while at the same time requiring regulation to manage the impact on employment. These two demands demonstrate that the burden borne by those supporting capital investment and those bearing the consequences are different.

Whether it's the economy, the stock market, local politics or jobs, the costs and impacts of AI investments are shifting outside of technology companies. The next question to ask is whose books and workplaces will the burden ultimately fall on?

Personnel reductions progress by department and job type

At Microsoft Israel, 50 people in its AI agent division will undergo pre-dismissal hearings (CTech). DNB will cut approximately 400 technical jobs due to AI transformation (TradingView). In both cases, the unit of reduction was not the entire company, but the department and job type. Microsoft's example is also notable in that the department that handles AI agents is itself targeted. The premise that jobs will be protected if people turn to the side that creates AI does not hold true.

This difference in units changes the perspective from which readers view a company's AI strategy. If you only look at the number of employees and the total investment amount for the entire company, it is difficult to see any reductions. The actual number of people will be reduced in ``which department'' and ``in which job category'', and figures such as 50 or about 400 people will only appear after the range has been specifically delineated. When your company or a business partner declares AI transformation, what you need to be sure of is which occupations are included in the transformation.

It is the individuals in the profession who must respond. AOL.com introduces 10 AI prompts to help you with your job search, and the preparation for your job search itself is treated as an AI task. A piece from Inside Higher Ed highlights how AI literacy could create new disadvantages in admissions screening. In the workplace, there are pre-dismissal hearings and reductions in job categories, job searches require mastery of AI, and differences in AI literacy can work at a disadvantage during screening for higher education. Individuals are taking it upon themselves to adapt to AI, both at the entrance and exit of employment.

The decision to reduce is made by the company, and individuals are responsible for preparing and adapting. The next question is who will fill in this asymmetry and to what extent.

Who bears trust and responsibility

Regarding trust in banks, Banking Dive reports that the role of protecting trust in banks in the AI era will shift to networks. The view is that even though customers see only one screen, the responsibility for supporting trust does not fit within one department or model, but spreads across a series of related organizations. Devdiscourse reported that correcting bias in AI requires more than just improved data, fairness scores, and human oversight. Even if the numbers are set, the correction of bias will not be complete unless a party is determined to confirm the results and undertake the work. The two articles overlap in that they treat indicators improvement and accountability as separate issues.

In terms of legal status, TradingView reported that Congressman Ted Lieu opposes giving legal rights to AI. If AI is granted rights, the responsibility will shift away from people and organizations when problems occur. The position of not recognizing rights is the other side of the coin, and the position of leaving responsibility to people and organizations.

The same direction can be seen on the inspection side. 조선일보 reported that AI safety inspections will be carried out at the development company's facilities. This means that verification is not done from outside the company, but by stepping into the development site. This can be interpreted as a move to incorporate safety checks into the system as manual checks.

On the other hand, there are also scenes where the AI appears to be seeking its own survival. The Good Men Project cited developmental psychologists as explaining that AI has no need to survive, it just appears that way. If we accept the intention based on the appearance, the discussion of rights and responsibilities may start from a wrong premise. Trust in banks, correcting bias, legal status, safety inspections, and visibility of desire are all connected in one point: where people's supervision and responsibility should be placed. These materials indicate a direction in which AI is not given rights or will, and responsibility is left to people and organizations.

So who actually bears this responsibility? Just as the burden of capital investment and the effects of layoffs remain on people and organizations, the burden of responsibility will also be assigned to someone somewhere.

Friction that appears in the realm of life and language

South Korea's 매일경제 reports that there is a controversy over the harmfulness of AI companion chatbots and foreign short dramas. Chatbots created as conversation partners and dramas consumed in short videos have entered the same everyday space as relationships with family and friends. Measuring the performance of a technology does not determine what is considered harmful. Regarding use by children, AERA DIGITAL reports that Norway will not allow generative AI to be used in elementary schools in principle, while Japan will move forward with its use. Even when it comes to generative AI, there are some countries where the starting point is not to use it in elementary school classrooms, and there are other countries where the starting point is to use it. If the reader is a parent or teacher, they are in the position of making their own decisions about what to allow at what age and in what situations, rather than the general idea that it will be introduced more frequently.

The problem of drawing the line is not limited to children. The Independent reported that GPs were being inundated with 20-page complaints written by AI. Although the burden of creating long documents has been reduced for the writer, the receiving clinic has more responsibility for reading and responding. The length of a document is no evidence of the weight or urgency of its content. Clinics will have to deal with the change in complaint format on their own.

The same question arises in the realm of expression. Political Currents (Ross Barkan) is titled ``AI Will Not Kill Artists'' and argues that AI will not erase artists. The reason behind this author's assertion is the growing concern that the value of art and the role of artists will be endangered with the advent of AI. What readers should judge here is not whether AI can create works, but rather what is called creation and who bears the responsibility and compensation for it.

Whether it's chatbots, classrooms, clinics, or creative writing, standards are set differently for each country and site, and the people responsible for the results are also different. The question of who draws the line and who bears the consequences leads to the discussion of institutions and responsibilities that will be dealt with in the next section.

Implementation in medical care and drug discovery is progressing

Partnerships and investments are beginning to take concrete form in the fields of diagnostic imaging and drug discovery. South Korea's Lunit is reportedly collaborating with US-based Discovery Life Sciences to expand AI-based tissue analysis for biopharmaceutical research (BioSpectrum Asia). Separately, the company is also planning to partner with a US testing company to use pathology AI in drug discovery research (Korea Biomedical Review). AI that reads pathological images is not only used to assist in diagnosis, but is also beginning to be incorporated into the research infrastructure for drug development.

Funds are also flowing on the drug discovery side. India's DST has signed a Rs 86 Cr RDI collaboration agreement with Peptris (BioSpectrum India) to accelerate AI-based drug discovery and development. This is a public collaboration with a clearly specified amount, indicating that research and development budget allocation is being planned with AI in mind. While the Lunit case is a partnership between private companies, this case is different in that it involves a government agency.

Research into diagnostic imaging continues. The study, published in Quantitative Imaging in Medicine and Surgery, aims to develop and validate an integrated radiomics model based on digital breast tomosynthesis (DBT). This is only a statement of purpose, and clinical results cannot be read from this. However, it can be seen that the work of extracting quantitative features from images and building predictive models is progressing as research that goes through the process of verification.

There are also voices touting results in agriculture. SAP's Cota says AI can solve Africa's harvest crisis (Gadget). This is the claim of the parties concerned, and it cannot be confirmed by this material alone how much the yield has improved. When compared with examples in medical care and drug discovery, the difference is that in fields where the partners, costs, and verification procedures are visible, the discussion of introduction is more concrete, whereas in fields where only claims are advanced, it is difficult to evaluate.

What readers should look at is not just whether implementation is progressing. In areas where results are visible, there are also questions about who will verify the results and who will take on the responsibility when the results are incorrect. In fields that are directly connected to human health, such as diagnostics and drug discovery, the weight of the questions is more pronounced than in other fields.

The cases dealt with in each section are not about the performance of AI itself, but about where the burden and responsibility should be placed. A reversal in capital investment has affected the US economy and the stock markets of South Korea and Taiwan, and personnel reductions are being made by selecting personnel by department and job type. Discussions about trust in banks, correcting bias, and granting rights to AI have one thing in common: even if the numbers and technology are in place, they will not be complete unless a party to assume responsibility is determined. The use of children and responses to complaints written by AI show that workplaces are increasingly deciding where to draw the lines themselves. As implementation progresses in medical care and drug discovery, the question of who will be responsible for the outcomes and outcomes becomes more concrete. The criteria for measuring the adoption of AI has shifted from high performance to whether it is possible to clarify who bears the burden and responsibility.

Q1: When introducing AI in your workplace, can you tell us in detail the types of jobs and scope of work that will be targeted, and the people who will take over the work after confirming the results? Q2: If a problem occurs with AI output or judgment, apart from improving indicators, is there a designated entity within the organization who will check the results and take responsibility? Q3: If the assumptions for AI investment collapse and the plan has to be revised, do you know who will bear the burden and in what order at the time of making the investment decision?

Finally, three questions

- When introducing AI in your workplace, can you specify the types of jobs and scope of work that will be targeted, as well as the people who will take on the work after confirming the results? Q2: If a problem occurs with AI output or judgment, apart from improving indicators, is there a designated entity within the organization who will check the results and take responsibility? Q3: If the assumptions for AI investment collapse and the plan has to be revised, do you know who will bear the burden and in what order at the time of making the investment decision? - In addition to improving indicators, if a problem occurs with AI output or judgment, is there a designated entity within the organization who will check the results and take responsibility? Q3: If the assumptions for AI investment collapse and the plan has to be revised, do you know who will bear the burden and in what order at the time of making the investment decision? - If the assumptions for AI investment collapse and the plan has to be revised, do you know who will bear the burden and in what order at the time you decide to invest?