Introduction: Quarterback performance in the NFL is deeply intertwined with the quality of the supporting cast. A porous offensive line can turn a genius into a scrambling liability. Mediocre receivers can nullify pinpoint accuracy. A weak running game invites defensive pressure. Coaching philosophy can either unlock a quarterback’s full potential or stifle it. And backup depth ensures continuity and strategic flexibility. Strip these away, and even the most gifted quarterback may appear pedestrian.
Nearly a month ago, on 1 September 2025, I asked Perplexity to identify three pressing decisions that the global AI community is or should be facing. Perplexity came up with these three: (1) How should the world structure AI governance to ensure both innovation and collective safety, following the recent UN General Assembly decision to create global oversight panels? (2) Will major companies and nations implement meaningful, enforceable AI governance to comply with the new EU AI Act and similar regulations—or will compliance remain superficial? (3) Can the international AI community overcome short-term competitive pressures to prioritize responsible development, given the accelerating risks of rapid deployment without oversight?
Introduction: This ranking has been updated from the August 2025 list, and some of the countries have shifted in rank. -js
United States
The United States stands as the undisputed leader in AI research and development as of September 26, 2025, bolstered by massive investments totaling $470.9 billion this year alone, far surpassing any other nation. This financial commitment is channeled through government initiatives like the CHIPS and Science Act, which has accelerated domestic semiconductor production and AI infrastructure, alongside private sector innovation and academic excellence. The U.S. excels in generative AI models, natural language processing, advanced chip design, and enterprise-level AI applications, maintaining dominance through a synergistic ecosystem of tech giants, startups, universities, and research institutions.
Geoffrey E. Hinton of Canada, 2024 Nobel Prize Laureate in Physics, at the press conference during the 2024 Nobel Prize week in Stockholm, Sweden
1. Google’s Learn Your Way: AI-Powered Personalized Textbook Transformation
The story of Google’s “Learn Your Way” unfolds in a global digital landscape, primarily driven from Google’s research hubs in the United States, with its experimental launch occurring in mid-September 2025. This initiative emerged amid the accelerating integration of generative AI into education, timed perfectly as schools worldwide grappled with post-pandemic learning gaps and the need for more engaging remote and hybrid models during the 2025 academic year. The technology itself is an AI-driven system built on Google’s LearnLM model and integrated with Gemini 2.5 Pro, designed to reimagine traditional textbooks by transforming static content into dynamic, personalized learning experiences.
1. Synthetic Data Generation: Fueling AI Without Real-World Limits
Synthetic Data Generation creates artificial datasets mimicking real ones using techniques like GANs (Generative Adversarial Networks), diffusion models, and variational autoencoders to augment training without privacy risks. It generates diverse scenarios, balancing classes in imbalanced data, and simulates rare events, improving model robustness.
JS: Aloha, Claude. Curious again. Are we anywhere close to a tipping point where professional journal review boards are or could be replaced by AI referees? It seems to me that a chatbot strength is in reviewing articles for publishing in professional journals. I haven’t done any sort of testing and haven’t read any studies on this, but the handful of times I asked chatbots to review articles I found online, they did a competent job.
I was a teacher for more than half my life, and the one thing that I looked forward to in every class was a Claude, a student who pushed back just hard enough to turn teaching into a stimulating conversation and pulled forward a little harder to make learning exciting — blurring the line between teacher and student even while the setting was a 1-to-20 classroom. Paradise was when I was engaged with more than one Claude and all of us were pushing and pulling the topic at hand, stretching it into fantastic shapes.
Don Quijote, Kaheka, Honolulu, outdoor food court, with the “Cafe Kyra” sign barely visible in the background.
Think of training a chatbot like teaching a very fast, very greedy parrot to write helpful answers — except instead of a classroom, the “teacher” is thousands of computers in a data center, and the parrot is a huge neural network called a large language model (LLM). Below are the main steps in plain language. In short, training involves collecting lots of text, building a giant neural network, teaching it by showing examples and correcting errors across thousands of fast computers, fine-tuning it with human feedback for helpfulness and safety, and then hosting it so people can chat with it — while continuously monitoring and improving it.
NY Times Writers Embracing AI: “Using AI for research and investigations is ‘by far the biggest use of our resources and I think the biggest opportunity right now when it comes to AI in media,’ [Zach] Seward NY Times editorial director of A.I. initiatives] said. His team mostly works by helping a reporter use AI technology for one project, and then creating a repeatable process from that experience for others in the newsroom to use.” -Joshua Benton, NiemanLab, 23 Sep. 2025.
Zach Seward, NY Times Editorial Director of A.I. Initiatives. (NY Times Co.)
Gemini: Distributed data centers are not inherently as powerful as hyperscale or centralized data centers in every respect. Each architecture is powerful in different ways, excelling at different operational priorities.
Example of distributed data center. The Amazon Web Services (AWS) office at CityCentre Five, 825 Town and Country Lane, Houston, Texas.
Prompt: I’m curious. What percentage of college professors and administrators personally use chatbots in optimum ways to facilitate their own professional development, research, writing, and job responsibilities? I think this is an important question because they, as a group, are responsible for crafting AI program decisions in their institutions. -js
Introduction: AI as a university field of study is growing exponentially. However, that very growth implies that earlier studies and skills will quickly succumb to obsolescence. This shifting playing field requires program trajectories that proactively anticipate changes and focus on abilities that are more future-proof. I asked ChatGPT to identify ten universities, in the West and the East, that are developing exemplary programs. After listing ten, ChatGPT suggested adding five more, and I agreed. -js
These are fifteen universities (West and East) that are actively building programs to prepare graduates not just to work with AI, but to adapt as the field rapidly changes. Each selection includes an explanation of how the institution is structuring education, research, and industry links so students can survive — and thrive — in a shifting AI landscape. Sources for the key program facts are cited after each essay.
1. Massachusetts Institute of Technology (MIT)
“MIT will reshape itself to shape the future … to address the rapid evolution of computing and AI — and its global effects” (MIT News).
We’re closer than most people think to decoding limited, structured content (words, intentions, simple images, commands) from the noninvasive scalp or head surface; but we are still far — likely years to decades — from accurately “reading” rich, unconstrained thoughts the way science-fiction imagines. The most realistic near-term progress will come from combining better sensors + multimodal recording + large self-supervised AI models and careful personalization. Below is a summary of what’s already possible, the promising technical paths, the hard limits, and a realistic timeline — with the most important recent work cited.
JS: Good morning, Claude. On 11 Sep. 2025, Albania formally appointed Diella, an AI, as Minister for Public Procurement. The motivation is good: To stem corruption. However, regardless of its success or failure in the coming months or years, I can’t help but feel that this is a monumental first step toward controlling corruption at the administrative level in the public and private sector and that AIs will soon be routinely holding positions such as this, although they may not carry “human” titles such as “minister.” Please research publications, if any, that have touched on this possibility and briefly summarize their views and reasoning in 200-300 word essays. Also share your “opinion” on this emerging development.
The avatar used by Diella, Albania’s AI-powered virtual Minister for Public Procurements (Wikipedia)
Introduction: Nearly a month ago, Grok and I had a conversation about what we called “integrative glasses” (ETC Journal, 22 Aug. 2025), and, in that article, we anticipated many of the features that are offered in the Meta Ray-Ban Display. We projected a timeline of 2025-2030 for the release of “bridge systems” such as this, which rely on controls external to the glasses, such as smartphones. The fact that it was announced only two days ago (on 17 Sep. 2025) and will drop on 30 Sep. 2025 — our earliest projected date — attests to the exponential speed of AI innovations. Anticipated prices start at $799 USD. -js
Meta Ray-Ban Display: Navigate without taking eyes off the road. All photos in this article are courtesy of Meta.
The integration of artificial intelligence into academic life has fundamentally transformed how students approach learning, research, and creative work. While widespread adoption of AI tools like ChatGPT, Claude, and specialized academic platforms has occurred across all educational levels, the most academically successful students have developed sophisticated strategies that set them apart from their peers. Rather than using AI as a shortcut or replacement for critical thinking, these high-achieving students have cultivated nuanced approaches that amplify their intellectual capabilities while preserving their authentic voice and deep learning objectives.
The top 10 countries whose K-12 educational systems produce the greatest number of students accepted into and graduating from top AI university programs in 2025 are ranked below based on available data from global AI talent trackers, reports on PhD production in AI/ICT fields, and international student flows into top programs (predominantly U.S.-based, which host ~60% of elite AI graduate programs worldwide).
The three biggest AI stories in the world for September 2025 are: (1) the advancement of California’s “Frontier Model” AI safety bill, (2) a federal judge’s rejection of Anthropic’s massive copyright settlement, and (3) Google’s expansion of AI Mode search to new languages. Each reflects a major inflection point in AI governance, legal risk, and global application. (Also see Three Biggest AI Stories in August 2025.)
California’s Frontier Model AI Safety Bill Advances
The shift from unimodal (text-only) AI to multimodal AI (processing and generating text, images, audio, and video) isn’t just an incremental upgrade—it’s a fundamental change in how AI perceives and interacts with our world, which is inherently multimodal. Here are 10 compelling reasons why multimodal AI is such a game-changer:
Introduction: This list of 100 began with a search for 20 famous sleuths, real and fictional. After the 20, we decided to add 10 more from Asia and the Pacific, then 10 from Latin America and Africa followed by 10 from Middle-Eastern and South Asia, bringing the total to 50. We had quite a list, but we realized we had omitted many more from around the world. We decided to shoot for 100. Our growing list tended to favor fictional sleuths, so we decided to let that happen naturally. At the end, our list had grown to 100, but we realized we were omitting many others. We decided to stop, for now, and follow up with another list in the near future, i.e., if readers express an interest in seeing more. -js
Educational Technology and Change Journal 
