Stefaan Verhulst

Report by OpenAI: “More than 5% of all ChatGPT messages globally are about healthcare, averaging billions of messages each week. Of our more than 800 million regular users, one in four submits a prompt about healthcare every week. More than 40 million turn to ChatGPT every day with healthcare questions.
In the United States, the healthcare system is a long-standing and worsening pain point for many. Gallup finds that views of US healthcare quality have sunk to a 24-year-low; that Americans give the system a C+ on access and a D+ on costs; and that a combined 70% believe the system has major problems or is in a state of crisis. In our own research, three in five Americans say the current system is broken, and strong majorities tell us that hospital costs (87%), poor healthcare access (77%), and a lack of nurses (75%) are all serious problems.
For both patients and providers in the US, ChatGPT has become an important ally, helping people navigate the healthcare system, enabling them to self-advocate, and supporting both patients and providers for better health outcomes.

Based on anonymized ChatGPT message data:
– Nearly 2 million messages per week focus on health insurance, including for comparing plans, understanding prices, handling claims and billing, eligibility and enrollment, and coverage and cost-sharing details.
– In underserved rural communities, users send an average of nearly 600,000 healthcare-related messages every week.
– And seven in 10 healthcare conversations in ChatGPT happen outside of normal clinic hours.

This report details: (1) how users are turning to ChatGPT for help in navigating the US healthcare system; (2) how they’re turning to ChatGPT to help them close healthcare access gaps, including in “hospital deserts” across the country; and (3) how healthcare providers and workers are using AI in their roles now…(More)”.

Paper by Hangcheng Zhao and Ron Berman: “Large language models (LLMs) change how consumers acquire information online; their bots also crawl news publishers’ websites for training data and to answer consumer queries; and they provide tools that can lower the cost of content creation. These changes lead to predictions of adverse impact on news publishers in the form of lowered consumer demand, reduced demand for newsroom employees, and an increase in news “slop.” Consequently, some publishers strategically responded by blocking LLM access to their websites using the robots.txt
file standard.
Using high-frequency granular data, we document four effects related to the predicted shifts in news publishing following the introduction of generative AI (GenAI). First, we find a consistent and moderate decline in traffic to news publishers occurring after August 2024. Second, using a difference-in-differences approach, we find that blocking GenAI bots can have adverse effects on large publishers by reducing total website traffic by 23% and real consumer traffic by 14% compared to not blocking. Third, on the hiring side, we do not find evidence that LLMs are replacing editorial or content-production jobs yet. The share of new editorial and contentproduction job listings increases over time. Fourth, regarding content production, we find no evidence that large publishers increased text volume; instead, they significantly increased rich content and use more advertising and targeting technologies.
Together, these findings provide early evidence of some unforeseen impacts of the introduction of LLMs on news production and consumption…(More)”.

Whitepaper by Frontiers: “…shows that AI has rapidly become part of everyday peer review, with 53% of reviewers now using AI tools. The findings in Unlocking AI’s untapped potential: responsible innovation in research and publishing point to a pivotal moment for research publishing. Adoption is accelerating and the opportunity now is to translate this momentum into stronger, more transparent, and more equitable research practices as demonstrated in Frontiers’ policy outlines.

Drawing on insights from 1,645 active researchers worldwide, the whitepaper identifies a global community eager to use AI confidently and responsibly. While many reviewers currently rely on AI for drafting reports or summarizing findings, the report highlights significant untapped potential for AI to support rigor, reproducibility, and deeper methodological insight.

The study shows broad enthusiasm for using AI more effectively, especially among early-career researchers (87% adoption) and in rapidly growing research regions such as China (77%) and Africa (66%). Researchers in all regions see clear benefits, from reducing workload to improving communication, and many express a desire for clear, consistent policy recommendations that would enable more advanced use…(More)”.

Press Release by IBM: “Record-setting wildfires across Bolivia last year scorched an area the size of Greece, displacing thousands of people and leading to widespread loss of crops and livestock. The cause of the fires was attributed to land clearing, pasture burning, and a severe drought during what was Earth’s warmest year on record.

The Bolivia wildfires are just one, among hundreds, of extreme flood and wildfire events captured in a new global, multi-modal dataset called ImpactMesh, open-sourced this week by IBM Research in Europe and the European Space Agency (ESA). The dataset is also multi-temporal, meaning it features before-and-after snapshots of flooded or fire-scorched areas. The footage was captured by the Copernicus Sentinel-1 and Sentinel-2 Earth-orbiting satellites over the last decade.

To provide a clearer picture of landscape-level changes, each of the extreme events in the dataset is represented by three types of observations — optical images, radar images, and an elevation map of the impacted area. When storm clouds and smoky fires block optical sensors from seeing the extent of flood and wildfires from space, radar images and the altitude of the terrain can help to reveal the severity of what just happened…(More)”.

Article by Louis Menand: “Once, every middle-class home had a piano and a dictionary. The purpose of the piano was to be able to listen to music before phonographs were available and affordable. Later on, it was to torture young persons by insisting that they learn to do something few people do well. The purpose of the dictionary was to settle intra-family disputes over the spelling of words like “camaraderie” and “sesquipedalian,” or over the correct pronunciation of “puttee.” (Dad wasn’t always right!) Also, it was sometimes useful for doing homework or playing Scrabble.

This was the state of the world not that long ago. In the late nineteen-eighties, Merriam-Webster’s Collegiate Dictionary was on the Times best-seller list for a hundred and fifty-five consecutive weeks. Fifty-seven million copies were sold, a number believed to be second only, in this country, to sales of the Bible. (The No. 1 print dictionary in the world is the Chinese-language Xinhua Dictionary; more than five hundred million copies have sold since it was introduced, in 1953.)

There was good money in the word business. Then came the internet and, with it, ready-to-hand answers to all questions lexical. If you are writing on a computer, it’s almost impossible to misspell a word anymore. It’s hard even to misplace a comma, although students do manage it. And, if you run across an unfamiliar word, you can type it into your browser and get a list of websites with information about it, often way more than you want or need. Like the rest of the analog world, legacy dictionaries have had to adapt or perish. Stefan Fatsis’s “Unabridged: The Thrill of (and Threat to) the Modern Dictionary” (Atlantic Monthly Press) is a good-natured and sympathetic account of what seems to be a losing struggle…(More)”.

Book by C. Thi Nguyen: “…takes us deep into the heart of games, and into the depths of bureaucracy, to see how scoring systems shape our desires.

Games are the most important art form of our era. They embody the spirit of free play. They show us the subtle beauty of action everywhere in life in video games, sports, and boardgames—but also cooking, gardening, fly-fishing, and running. They remind us that it isn’t always about outcomes, but about how glorious it feels to be doing the thing. And the scoring systems help get us there, by giving us new goals to try on.

Scoring systems are also at the center of our corporations and bureaucracies—in the form of metrics and rankings. They tell us exactly how to measure our success. They encourage us to outsource our values to an external authority. And they push on us to value simple, countable things. Metrics don’t capture what really matters; they only capture what’s easy to measure. The price of that clarity is  our independence.

The Score asks us is this the game you really want to be playing?…(More)”.

Crust News: “At some point this year it became obvious that simply writing about immigration enforcement in the United States was no longer enough. Every time something happened, it happened in isolation. A raid on Canal Street, an abduction in Chicago, an ICE agent that made a single headline for his actions. But the reality of authoritarianism, is that it is an entire system, there are no isolated incidents, they are all connected.

There was nowhere to see these connections in their entirety, so we built a place for it to stay. And best of all, we’re doing so outside the USA, where Trump’s regime can’t get to us.

The ICE List Wiki is now public. It documents immigration enforcement activity across the United States, not just ICE, but Border Patrol, HSI, DHS more broadly, and the hundreds of local police departments operating under 287(g) agreements. Agent identities, incidents, raids, vehicles, supporting agencies, and companies propping up the regime, are recorded as the interconnected system that they are. Entries and linked to each other so that nothing exists on its own anymore. This is our Christmas gift to the USA: a record that refuses to forget. The reason this became necessary has everything to do with the political moment we are in. Trump’s return to power has accelerated an enforcement machine that was already dangerous, but not yet authoritarian, but ripe to become so. What exists now is a system that moves quick, loud, and with very little interest in being legible to the public, avoiding accountability at every step.

Authoritarianism doesn’t usually arrive with a single dramatic act. It arrives through administration, repetition, and exhaustion. They break you down, and you forget how bad those initial steps were, because they had become normalised.

There is a huge misconception that this is just ICE, it’s not. ICE shows up, but so does CBP, HSI, DEA, FBI, local police, and even postmasters. 287(g) agreements turn police officers into extensions of Trump’s extremism while allowing everyone involved to hide behind the headlines about ICE, as if that is all that is going wrong in this moment. Together, these corrupted organisations are forming something much larger, much darker, and much more frightening than anything the USA has seen at home, but reminiscent of what the USA has seen in historical wars abroad.

We want to remove the misconception and track the whole thing. As much as we possibly can…(More)”.

Article by Shana Lynch: “…After years of fast expansion and billion-dollar bets, 2026 may mark the moment artificial intelligence confronts its actual utility. In their predictions for the next year, Stanford faculty across computer science, medicine, law, and economics converge on a striking theme: The era of AI evangelism is giving way to an era of AI evaluation. Whether it’s standardized benchmarks for legal reasoning, real-time dashboards tracking labor displacement, or clinical frameworks for vetting the flood of medical AI startups, the coming year demands rigor over hype. The question is no longer “Can AI do this?” but “How well, at what cost, and for whom?”

Learn more about what Stanford HAI faculty expect in the new year…As the buzz around the use of GenAI builds, the creators of the technologies will get frustrated with the long decision cycles at health systems and begin going directly to the user in the form of applications that are made available for “free” to end users. Consider, for example, efforts such as literature summaries by OpenEvidence and on-demand answers to clinical questions by AtroposHealth. 

On the technology side, we will see a rise in generative transformers that have the potential to forecast diagnoses, treatment response, or disease progression without needing any task-specific labels.

Given this rise in available solutions, the need for patients to know the basis on which AI “help” is being provided will become crucial (see my prior commentary on this). The ability for researchers to keep up with technology developments via good benchmarking will be stretched thin, even if it is widely recognized to be important. And we will see a rise in solutions that empower patients to have agency in their own care (e.g., this example involving cancer treatment)…(More)”.

Book by Ben Zweig: “…offers a revolutionary approach to transforming human capital management through the power of taxonomies. The book follows the experience and ideas of key individuals―from the founders of Wall Street, to the original management consultant, to a young data scientist just out of grad school looking to make sense of the modern workforce―in order to illustrate why our current human capital infrastructure is not serving employees well and what we can do to change that.

By categorizing and organizing workforce data, Zweig provides a practical roadmap for creating a more efficient and data-driven labor market. This book includes key insights on how to:

• Use AI and similar large language model technologies to support businesses with appropriate categorization and regimentation of data
• Know whether or not a taxonomy can be useful and functional for an organization in their ability to be flexible, auditable, and adaptable
• Build a taxonomy that meets the needs of a workforce or organization through clustering, labeling, and production

Combining storytelling with real-world examples, theoretical analysis, and a practical framework, Job Architecture is an essential guide for companies to manage a competitive, modern workforce that improves the working experience for all employees…(More)”.

World Bank Report: “This brief presents the 2025 update of the GovTech Maturity Index (GTMI), offering a global snapshot of public sector digital transformation across 197 economies. The GTMI assesses four focus areas, Core Government Systems (CGSI), Online Public Service Delivery (PSDI), Digital Citizen Engagement (DCEI), and GovTech Enablers (GTEI), using 48 indicators. The methodology combines self-reported survey data from 158 economies with publicly available information for the remaining 39. Findings indicate overall progress since 2022 but widening disparities between higher-income (Group A) and lower-income (Group D) economies. Advances are noted in core systems (e.g., government cloud) and service delivery (e.g., customs services, digital ID), while digital citizen engagement remains the least mature area and adoption of a whole-of-government approach is limited. The brief recommends accelerating implementation of interoperability frameworks, strengthening sustainability of online service portals, and updating GovTech strategies in line with evolving technologies. It underscores the need for targeted support to low-income regions, particularly in Africa, and calls for clear monitoring frameworks to track progress and inform evidence-based policymaking…(More)”.