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

AI Highlights — the whole picture — 2026-11-20 (Fri) Daily News

Daily ReportDaily edition, 03:10 JST

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 Report03:47 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  ·  33 articles
AI INTEGRATED ANALYSIS2026-11-20 (Fri) — Daily Report · 03:47 JST
Winners determined by computing resources and trust

Isomorphic Labs added Merck to its partnership just days after raising $5 billion, and Accenture is reportedly expected to cut about 950 jobs in Portugal around the same time. While the results are concentrated on the side with the funds and computational resources, the burden of employment, finance, and social trust remains on the other side. Growth rate numbers alone do not reveal who is shouldering the burden and how much. When reading today's article, rather than looking at how much growth the introduction of AI has brought about, I would like you to look at where the burden remains. The facts in each verse are connected at this one point.

Funds and computational resources become a moat for drug discovery

Isomorphic Labs has added Merck to its AI drug discovery partnership (Fierce Biotech), just days after raising $5 billion. The target of the partnership is cancer drug discovery using AI, and is reported to be a joint effort with MSD (Yahoo Finance, Unite.AI). During the same period, Boltz raised $110 million to expand its AI tools for drug discovery (The Pharma Letter). Companies that have raised large amounts of funding are also forming alliances with major pharmaceutical companies. When readers compare drug discovery AI companies, they are no longer looking at the newness of the technology, but rather on the depth of their funding and the lineup of their partners.

Geodesic's Gu Quanquan said it consumed $12 million in computing resources in six months. He said the advantage of AI drug discovery lies in GPUs, not wet labs (BigGo Finance). If this view is correct, the competitiveness of drug discovery AI will depend less on laboratory equipment and more on the financial strength to continue purchasing computing resources and the network of alliances that support it. The scale of $12 million in six months also shows that the amount of funding is linked to the consumption of computational resources. However, it is important to keep in mind that this is the view of one company's management.

The movements of major players are also similar to this trend. Novartis expects sales growth of 5% annually through 2031 and has raised peak sales targets for Kesimpta, Scemblix and Leqvio (MLQ.ai). Oracle announced an AI agent to coordinate clinical trial operations (Unite.AI). The former is the growth outlook of existing major companies, and the latter is the movement of AI into the field of test management. Both can be read as evidence that the introduction of AI and the outlook for profits are progressing among those with the funds and infrastructure.

If the results are concentrated on those who hold the computing resources and alliances, then the scale of growth alone will no longer be enough to measure adoption. We need to look at who is responsible for what outside of that growth. In the next section, we'll look at where that burden manifests itself.

Job cuts due to economic boom and human resource development being tested

Accenture is expected to cut around 950 jobs in Portugal, reports shattered.io. The headline is Meta-related, which indicates that the placement of personnel is in question for large-scale AI-related projects (shattered.io). Salesforce has also cut jobs, its stock price has plummeted, and its acquisition of Braze is in question, according to The American Bazaar. Meanwhile, Malaysia's GDP in the July-September period exceeded expectations, and Reuters reports that the AI boom will continue to boost growth in 2027. During the same period, growth numbers and layoffs numbers appear side by side.

What this arrangement shows is that looking at the growth rate of a country or company does not tell us what the footfall of the people working there is. Even if Malaysia's GDP improves, it won't offset the roughly 950 job cuts at Accenture and the cuts at Salesforce. Those evaluating the introduction of the system need to look at the growth figures and then separately confirm which regions and which occupations are experiencing reductions.

Behind the scenes of reductions, there is also a movement to train people. Instep has launched an apprenticeship program to develop business personnel who are strong in AI (HR News). BDO holds its first company-wide training week, highlighting its focus on human skills alongside AI (Consultancy.uk). However, in the scope of both articles, it is unclear whether the scale of training is commensurate with the scale of reductions. The reduction of approximately 950 jobs and the apprenticeship program and one training week simply cannot be compared. If human resource development is to be included in the evaluation of its introduction, it is necessary to look at not only whether it has been implemented, but also the extent to which there is a path for those who have been retrenched to work again.

Seen in this way, the benefits of AI will be concentrated in those who control computing resources and partnerships, while the coordination and development of personnel will fall to individual companies and workers. The question of who will bear the burden extends not only to employment, but also to the mechanisms used to move funds.

Finance, regulation, and geopolitics will determine the speed of implementation

Forbes reported that the Fed's concerns about AI-related bank runs are becoming a reality (Forbes). TradingView reports that while Trump heads for talks with China, obstacles remain in the AI supply chain (TradingView). The former introduces uncertainty from the financial side, and the latter from the parts and supply chain side, which introduces uncertainty into the funding and procurement that is a prerequisite for the introduction of AI. All of these are facts stated in the headlines, and the magnitude of the impact needs to be confirmed in the main text of each article. Still, for those evaluating the implementation, in addition to internal cost-effectiveness, the financing environment and the geography of the procurement source will be considered as prerequisites for planning.

The same structure exists in the pharmaceutical field. BioSpace reported that Pfizer's CEO appealed for cost reduction through AI in the shadow of China's rise (BioSpace). The fact that calls for cost reduction are being made under pressure from the rise of China suggests that the introduction of AI is starting to be talked about more as a competitive imperative than as an efficiency option. STAT News' Pharmalittle covered Merck's AI partnership and topics surrounding China and the pharmaceutical supply chain in the same list (STAT News). The very fact that talk of alliances and supply networks are read side by side shows that the selection of partners for introduction is becoming inseparable from geopolitics. When reading an announcement of a partnership, readers need to ask not only about the computing resources of the partner they are partnering with, but also which regions their future supply networks will depend on.

On the regulatory side, Nature covered the impact of proposed changes to the EU Medical Devices Regulation on AI-powered devices (Nature). Because the content of the amendments relates to the approval process for each device, the speed of introduction will increasingly depend on how the rules are rewritten rather than the maturity of the technology. The extent of the impact cannot be determined without checking the final text of the proposed amendment, but when estimating the timing of introduction of AI-equipped devices, it is necessary to take the timing of the revision of the regulations into consideration.

Finance, supply chains, and regulations all determine the speed of adoption from outside the company. If this is the case, the next question is who will bear the losses and burdens that arise after the speed is determined.

Clinical research in the field progresses from skepticism to practical verification

In clinical and research settings, the debate surrounding AI is shifting from skepticism to practical verification. Clinical Lab Products tells the story of digital pathology's long journey from skepticism to AI. A survey of machine learning in rice cultivation introduced by Bioengineer.org covered 343 studies and showed that adoption is uneven depending on the field. BioMed research international also published a study that used bioinformatics and machine learning to identify key genes involved in mitochondrial homeostasis and immune cell infiltration in MASLD. It is one thing to have widespread adoption and another thing to have uniform usage everywhere. Readers need to check not only the reports of ``use'' but also the location of bias.

In the cancer field, there are many studies with clear verification designs. The BMC cancer study was a multicenter, retrospective study that tested a machine learning model that distinguishes between malignant and non-malignant pleural effusions in lung cancer patients. Naunyn-Schmiedeberg's archives of pharmacology includes a machine learning pipeline that searches for secreted protein biomarkers for colorectal cancer. It is unique in that it handles everything from single-cell transcriptomics to functional verification in a single process. A study in the Asian journal of andrology retrospectively predicted ISUP grade increase from biopsy to radical prostatectomy in an interpretable model that incorporated mpMRI-diagnosed prostatitis.

Research using images and molecular data continues. Biomedical physics & engineering express reported lightweight and interpretable brain tumor segmentation from multimodal MRI using supervoxel graph Transformer. Cancer Cell International's research has created a decision-making system for determining distant metastasis of mixed hepatocellular carcinoma/cholangiocarcinoma using a retrospective cohort using machine learning. Molecular & cellular proteomics (MCP) featured a study that used mass spectrometry and machine learning to classify renal cell carcinoma and benign tumors and analyze their molecular phenotypes. What they all have in common is that their titles indicate a multi-institutional and retrospective verification framework and consideration for interpretability. Those evaluating the introduction will not only ask about accuracy figures, but also which groups and which procedures were used to confirm the results.

The more verifications are carried out in the field, the more specific the questions become about the time it takes to verify results and who will take responsibility for incorrect decisions. It is from this stage that the evaluation scale shifts to the destination of the burden.

Questions of distance, bias and loneliness between humans and AI

The distance between young people and AI is already being measured in their daily lives. A Gallup survey found that 1 in 3 young people considered AI a friend, and the same survey also showed high levels of loneliness (Martin Cid Magazine). Many users were also concerned about the report that LGBTQ+ young people were seeking mental health advice from an AI chatbot (GO Magazine). The satirical publication The Onion depicts how people spend more time with their AI girlfriends and AI personal trainers (The Onion). Although this is satire rather than factual reporting, it does show that we have entered the stage where the increasing amount of time people spend with AI has become a laughing matter. For those evaluating the introduction, an increase in the usage rate is not only an indicator of success, but also an indicator of the burden on those who are at risk of losing their counselors.

The discussion of bias and awareness influences how that burden is viewed. Vital City argues that if the data is biased, AI will amplify the problem (Vital City). The more young people use AI for advice, the more biased learning data will have a direct impact on how individual concerns are handled. Meanwhile, philosopher Nagel told WIRED that the idea that AI is conscious is "ridiculous." From this perspective, AI should be treated not as an opponent, but as a product for which designers and providers are responsible. The base of creators is also expanding. University of Milwaukee (UW-Milwaukee) student placed third in AI hackathon (UW-Milwaukee). AI has become a familiar tool not only for the young people who use it, but also for the young people who create it.

In the field of music, providers have begun to draw the line between distances. SoundCloud cracked down on adulterated songs mass-produced using AI, and Spotify expanded Artist Profile Protection (Music Ally). Both measures are aimed at distinguishing between things created by AI and things created by humans, and protecting the profile of the person in question. AI as a friend, AI as a consultant, and AI as a creator of mass-produced songs all come down to the same question: how to maintain distance from people. The next section deals with who decides the distance and who bears the cost.

The tendency for alliances and results to be concentrated in companies with financial and computational resources was confirmed in both articles on drug discovery and on the growth of the country as a whole. At the same time, the burdens of job cuts, uncertainties in finance and supply chains, uneven use of the field, and isolation among young people as they become more exposed to AI were felt elsewhere than on those who reaped the benefits. If we only look at the size of growth in the evaluation of implementation, these burdens will not be included in the figures. After confirming the magnitude of growth, we should also consider with equal weight which people, which organizations, and which social strata the burden is being transferred to. Evaluation criteria are shifting to this order.

Q1: Who in the field is affected by the increase or decrease in work and responsibilities caused by the introduction of AI? Q2: What records and procedures should be used to confirm whether bias or damage has been overlooked behind reports of widespread use? Q3: Do the indicators for measuring the success of the implementation include the burden borne by others, such as reductions and uncertainties?

Finally, three questions

- Who in the field is affected by the increase or decrease in work and responsibilities caused by the introduction of AI? Q2: What records and procedures should be used to confirm whether bias or damage has been overlooked behind reports of widespread use? Q3: Do the indicators for measuring the success of the implementation include the burden borne by others, such as reductions and uncertainties? - What records and procedures should be used to confirm whether bias or damage has been overlooked behind reports of widespread use? Q3: Do the indicators for measuring the success of the implementation include the burden borne by others, such as reductions and uncertainties? - Do the metrics used to measure the success of the implementation include any burdens assumed by others, such as reductions or uncertainties?

📚 Sources (all material)

Every item this issue drew on. External links open in a new tab. 33 items.

AI in Medicine (Cancer Care, Rare Diseases)

  1. Article (Clinical Lab Products) — Japanese-language summary

AI Regulation, Safety and Geopolitics

  1. Article (Nature) — Japanese-language summary
← 2026-11-19-morningIndex2026-11-21-morning →
← AI Highlights — the whole picture Index