AI Highlights — the whole picture — 2026-10-18 (Sun) Evening News

Growth predictions and investment enthusiasm for AI are already being expressed on a concrete economic scale. Anthropic's Clark said AI could significantly boost U.S. GDP. However, while there are examples of employees in the gambling industry reporting layoffs due to AI, some believe that mass layoffs are not yet visible across the economy. If you stand on the side of expectations, the changes will appear to be large, but if you stand on the side of the actual situation on the ground, the changes will appear to be limited. All of the topics gathered over the past half day fall into the same crossroads. Should we base our judgment on the growth that is expected to occur in the future, or on the employment and counseling opportunities that can be confirmed now? In this article, we will look at the differences in standards from four different perspectives: investment, personnel, lifestyle consultation, and professional settings.

Expectations for growth and the weight of the investment cycle

Anthropic's Clark said AI is advancing rapidly and could significantly boost U.S. GDP (Reuters). This indicates that growth prospects are beginning to be discussed on a concrete economic scale. Meanwhile, US bank BofA notes that the AI investment cycle could come under pressure from competition and falling prices (Yahoo Finance). The greater the expectations for growth, the more the capital investment to support them will depend on the prospects for recovery. Even if performance improves, if the price falls, the return on investment will change. Although the two statements seem contradictory, they are based on the same premise: that AI will continue to improve. The difference lies in whose profits come from the improvement.

The movement of capital and equipment is also geographically widespread. Nvidia plans to enter Europe, opening GTC Berlin from October 20th to 22nd (NVIDIA Blog). The event is being held just before the end of the conference, suggesting that interest in computing resources is shifting outside the United States. However, increasing the number of venues and seeing the investment bear fruit are two different things. The downward pressure on prices that BofA fears could similarly extend to its European base.

The involvement of the state is also becoming more visible. China is expanding its control over technology exports, from TikTok to AI models (Tomorrow's Affairs). If AI models are subject to export controls, the technology is subject not only to competition between companies but also to the conditions of permission for cross-border transfer. The US Department of Energy has released “Genesis Agent Skills” to support scientific research using AI (ExecutiveGov). This is not regulation, but the development of a public research infrastructure, and it can be seen that the government is involved in AI both in terms of restrictions and support. Private investment decisions will be made taking into account these national movements.

Looking at the materials so far, expectations are expressed in the large unit of GDP, investment is tied to competition and price pressure, and equipment and policies are tied to each country's circumstances. Those who make decisions based on growth projections assume that these conditions will hold in the future. So how are the numbers on the ground moving? It is necessary to check this alongside expectations, and in the next section we will turn our attention to the actual situation.

Disparity between personnel reduction and human resource development

The Times of Malta reports that employees in the gambling industry are reporting job cuts due to AI. The industry is narrowed down to the gambling industry, and reporting is done by front-line employees. In contrast, The Good Men Project discusses why AI has not yet caused mass layoffs. Both deal with the impact on personnel, but they look at things differently. The former refers to individual reductions that have occurred on the ground in specific industries, while the latter indicates the recognition that large-scale layoffs have not yet manifested in the economy as a whole. These two things can coexist. This is because it is possible for one industry to report cuts and for the economy as a whole not to have mass layoffs. Therefore, there is weak evidence to judge the overall trend based on just one report on reductions, or to discount changes at individual workplaces just by assuming that there are no mass layoffs overall.

There are also movements in a direction other than reduction. According to BigGo Finance, Mizuho Financial Group will train sales and administrative employees as AI human resources in three months. The targets are sales and clerical jobs, which are often thought to be easily replaced by AI. Instead of reducing the number of positions, companies are choosing to transform the same employees into human resources who can handle AI. Pasquale Pillitteri's 2026 Guide organizes Microsoft's free AI courses and certifications. Having an entryway to learning that does not incur tuition fees reduces the burden on individuals and companies when starting retraining. Mizuho FG's three-month period and Microsoft's free courses and certifications can be read as moves premised on retraining rather than cuts.

However, there are limits to what can be read from these materials. The gambling industry's report on reductions and the efforts by Mizuho FG and Microsoft to foster the same are two separate organizations and different industries. There is no evidence to suggest that reductions were replaced by training, and it is not possible to confirm whether or not the trained personnel were actually transferred. What the reader can use to judge is the fact that downsizing and retraining are occurring at the same time and in different places.

This discrepancy shows that the same event can be seen differently depending on the point in time in which the magnitude of change is measured. In the next section, we will look at the field of individual counseling outside of employment.

Dangers of AI becoming a consultant in daily life

ChatGPT has been sued for providing unqualified medical advice to a man who used it, causing a mental crisis (CBS News). This case shows that AI answers can be taken with the same weight as expert advice. Meanwhile, AOL.com has an article asking ChatGPT, "How can Gen Z become millionaires?" (AOL.com). Both medical care and asset building are areas where people should consult people with qualifications and experience. The ease of being able to do this in one chat eliminates the need to check the basis of an answer. What readers should judge is not only whether the AI's answer is correct, but also who is receiving the consultation in a manner that allows them to take responsibility for it.

The role of a counselor is also expanding into the private sphere. The Universidad de Córdoba (UCO) listing includes a service that features an "AI girlfriend" and offers voice conversations, as well as a service that provides chat with AI characters. The latter has a free image-to-video generator, features privacy controls, and is labeled as 18+ (Universidad de Córdoba (UCO)). These are used as substitutes for lovers or conversation partners, and users provide personal information such as their worries, preferences, and images. The very fact that a company claims to manage privacy indicates that the handling of information is a concern for users. However, it is not clear from the published content how far this management will actually function.

As these uses increase, the issue becomes more than just the safety of individual services. CT Examiner celebrates 50 years since the publication of ``The Selfish Gene,'' arguing that AI is testing humanity's future (CT Examiner). This means that the questions of what it means to be human and what people want from others are being posed to real-life users in consultations and intimate conversations with AI. The lawsuit can be read as an example of the first attempt to answer that question in the form of legal responsibility. The story so far shows that changes are already occurring in individuals' lives, independent of market expectations and employment statistics. The decision as to whether to use expectations or actual conditions as a standard depends on the extent to which such examples from the field are taken into account.

Identification of images and emotions in professional settings

IndexBox reports that the digital slide scanner market is expected to grow until 2035. The factors cited are remote pathology and AI boosts. Pathology is the work of converting specimens into slide images and reading them. If you use an image, you can separate the location of the inspection from the location of the reader. The ability to read the same images is also being tested in agriculture. According to Bioengineer.org, a new AI model has been reported that can detect crop diseases with high accuracy even in complex fields. A complex field is a real field where the background and lighting conditions are not perfect. The accuracy obtained under these conditions is more important as a basis for deciding whether to introduce the product than the accuracy obtained using well-equipped images in a laboratory. However, neither article shows how far implementation has progressed in the field. IndexBox tells you the market outlook, while Bioengineer.org tells you the detection accuracy of new models.

In the field of education, Bioengineer.org is communicating a new model that reads four faces simultaneously to determine students' emotions. Since it can handle multiple faces at once, it gets closer to the reality of the classroom than looking at each person one by one. However, the object of discrimination is human emotions, which have different characteristics than pathological specimens or crop leaves. In the field of pathology and crops, it is easy to check the results of diagnosis and detection later. Students' emotions make it difficult to set standards for checking whether their judgments are accurate. Even with the same "image reading" technology, the ease of implementation and the extent to which results can be trusted vary depending on the field.

Improvements continue to be made on the image creation side as well. TechRepublic introduced ways to improve the quality of AI images in ``6 Gemini Photo Editing Prompts for 2026.'' It can be seen that the user is at a stage where the user is improving the quality by devising the prompt. Although this is a different story from its introduction into professional settings, it can be seen that the ability to read and the ability to create are progressing side by side in the same image domain.

In other words, AI that reads images and facial expressions has already entered multiple industries. However, the progress of implementation varies by field. Market forecasts, research-stage models, and personal use innovations are not the same level of ``reality.'' It is necessary to decide for each field based on this difference whether expectations or reality should be used as the standard.

What all four sections have in common is that while the premise that AI capabilities will continue to improve is widely shared, the extent to which these improvements are reaching the field varies by domain. On the investment side, even if performance improves, if prices fall, the outlook for recovery will change. On the personnel side, the report of cuts in one industry and the absence of mass layoffs across the economy hold together. On the counseling side, in addition to the correctness of the answers, the question is whether the person receiving the consultation is in a manner that allows them to take responsibility. In professional settings, even if the market outlook and detection accuracy are shown, it is necessary to separately confirm how far implementation has progressed. Expectations and reality both partially reflect correctly. If you base your judgment on only one, your judgment on the other parts will be weakened.

Q1: Are you able to distinguish between the parts of your work that are actually handled by AI and the parts that are still handled by humans? Q2: Are there procedures in place for those receiving AI output to verify its basis and identify the parties responsible for errors? Q3: When making investment and hiring decisions, in what proportion do you rely on future growth prospects or current track records?

Finally, three questions

- Are you able to distinguish between the parts of your work that are actually handled by AI and the parts that are still handled by humans? Q2: Are there procedures in place for those receiving AI output to verify its basis and identify the parties responsible for errors? Q3: When making investment and hiring decisions, in what proportion do you rely on future growth prospects or current track records? - Are there procedures in place to enable those receiving the AI output to verify its basis and identify those responsible for errors? Q3: When making investment and hiring decisions, in what proportion do you rely on future growth prospects or current track records? - In what proportion do investment and hiring decisions depend on future growth prospects or current track records?