​(view in browser)​

Dear Reader,

Welcome to the 312th issue of our newsletter!

This week’s mix of ideas points to a maturing AI landscape where advantage comes less from raw access to models and more from the systems, habits, and interfaces built around them. On one side, there’s a clear push toward more capable and efficient infrastructure, from highly compressed models that promise sharper performance at lower cost to new frameworks that make it easier to connect models to tools and real workflows.

On another, the spotlight is shifting to practice: how developers structure local AI environments, how researchers can fold coding agents into academic work, and how specialized “skills” may help close the gap between what agents can do in theory and what they can reliably execute in production.

Running through all of it is a broader human theme: as AI spreads, the real differentiator may be the people who learn to work with it deliberately, whether that means designing systems for scientific discovery, building cleaner agent workflows, or developing the judgment and technical fluency that make automation an amplifier.

Our paper roundup sketches a moment in which AI is pushing simultaneously outward into society and inward toward the foundations of thought itself. Some of the work is strikingly concrete, showing how better statistical reasoning can reshape clinical trials, how large-scale behavioral data can reveal hidden structure in cities, and how careful population studies can cut through speculation in contested public-health debates.

At the same time, another thread probes the limits of machine understanding: whether systems that appear fluent actually see and reason in a meaningful way, whether generative AI should follow a different technological path, and whether questions about language, mind, and even consciousness, once reserved for philosophy, now have to be taken seriously as engineering concerns.

Our current book recommendation is "Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems" by A. Gullí. You can find all the previous book reviews on our website. In this week's video, we have a conversation between C. Newport and Y. LeCun on whether or not Are LLMs a Dead End.

Data shows that the best way for a newsletter to grow is by word of mouth, so if you think one of your friends or colleagues would enjoy this newsletter, go ahead and forward this email to them. This will help us spread the word!

Semper discentes,

The D4S Team

A. Gullí’s "Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems" feels like a timely guide for data scientists and machine learning engineers who are ready to move past the hype around AI agents and focus on how these systems are actually built. What makes the book stand out is its practical, pattern-based approach: instead of treating agents like magic, Gullí breaks them into reusable design ideas that help readers think more clearly about architecture, workflows, and implementation. That alone makes it more valuable than many AI books that are heavy on buzzwords and light on substance.

One of the book’s strongest qualities is its hands-on mindset. By working through recognizable frameworks and concrete design patterns, it gives technical readers a clearer path from experimentation to real system design. For ML engineers, that means a stronger grasp of modularity and maintainability; for data scientists, it offers a useful bridge between model knowledge and application building. The book is at its best when it helps readers see agentic systems not as mysterious novelties, but as engineering problems that can be approached systematically.

Its weaknesses are relatively minor but worth noting. Because it leans on current frameworks and tools, some parts may age quickly in such a fast-moving field, and readers looking for a deeper dive into evaluation, benchmarking, or production-scale operations may find it less comprehensive on those fronts. Still, Agentic Design Patterns sounds like the kind of book that can sharpen how technical practitioners think about intelligent systems—and for many readers, that will be reason enough to keep turning the pages.

1. ​Closing the knowledge gap with agent skills [developers.googleblog.com]
2. ​Anatomy of the .claude/ Folder [blog.dailydoseofds.com]
3. ​TurboQuant: Redefining AI efficiency with extreme compression [research.google]
4. ​Welcome to FastMCP [gofastmcp.com]
5. ​Designing AI for Disruptive Science [asimov.press]
6. ​What Young Workers Are Doing to AI-Proof Themselves [wsj.com]
7. ​Claude Code For Academics [github.com]
8. ​Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation [research.nvidia.com]

• ​A shared model-based linguistic space for transmitting our thoughts from brain to brain in natural conversations (Z. Zada, A. Goldstein, S. Michelmann, E. Simony, A. Price, L. Hasenfratz, E. Barham, A. Zadbood, W. Doyle, D. Friedman, P. Dugan, L. Melloni, S. Devore, A. Flinker, O. Devinsky, S. A. Nastase, U. Hasson)
• ​Embracing Bayesian Methods in Clinical Trials: FDA’s Long-Awaited Draft Guidance (J. J. Lee, F. E. Harrell Jr, L. M. LaVange, D. J. Spiegelhalter)
• ​Association between COVID-19 vaccination and sudden death in apparently healthy younger individuals: A population-based case-control study (H. Abdel-Qadir, H. A. Bhatt, S. Swayze, M. Paterson, D. T. Ko, D. N. Juurlink, J. C. Kwong)
• ​Validating urban scaling laws through mobile phone data: analysis of Brazil’s largest cities (R. S. Alencar, F. L. Ribeiro, H. Samaniego, R. Menezes, A. G. Evsukoff)
• ​MIRAGE: The Illusion of Visual Understanding (M. Asadi, J. W. O'Sullivan, F. Cao, T. Nedaee, K. Fardi, F.-F. Li, E. Adeli, E. Ashley)
• ​An Alternative Trajectory for Generative AI (M. Belova, Y. Kansal, Y. Liang, J. Xiao, N. K. Jha)
• ​Mathematical methods and human thought in the age of AI (T. Klowden, T. Tao)
• ​Could a Large Language Model be Conscious? (D. J. Chalmers)

Are LLMs a Dead End?

​

All the videos of the week are now available in our YouTube playlist.

​Unsubscribe​