​(view in browser)​

Dear Reader,

Welcome to the Juneteenth issue of the Data Science Briefing!

This week’s newsletter highlights several cutting-edge developments in the world of AI agents and security. A standout is the new paper, "Design Patterns for Securing LLM Agents against Prompt Injections," which introduces six actionable patterns—such as the Action-Selector and Plan-Then-Execute patterns—to help developers build large language model (LLM) agents that are robust against prompt injection attacks. These patterns offer a practical balance between agent utility and security, directly addressing the growing risks as LLMs are increasingly integrated with external tools and data sources. Complementing this, recent research on the automated design of agentic systems explores how meta-agents can autonomously generate and optimize agent architectures in code, potentially leading to more powerful and adaptable AI systems that outperform traditional hand-designed agents. Together, these advancements signal a shift toward both safer and more innovative agentic AI, underscoring the importance of security and automation in the next generation of intelligent systems.

Recent research continues to reveal the intricate interplay between social structures, technology, and human behavior. A large-scale study mapping face-to-face social networks in isolated Honduran villages demonstrates how understanding the fabric of personal relationships, such as trust and time spent together, can inform strategies for public health interventions and the spread of information. These findings are echoed in new epidemiological modeling work, where scalable Bayesian inference methods like ScITree leverage both genomic and social data to reconstruct transmission pathways, underscoring the value of detailed network knowledge in managing disease outbreaks. Meanwhile, as AI systems become more integrated into daily life, evidence is mounting that clinical knowledge encoded in large language models does not seamlessly translate to effective human interactions, highlighting a persistent gap between technical proficiency and social competence. Together, these studies emphasize the importance of both mapping and understanding human networks—whether for health, technology, or communication—and the ongoing challenge of ensuring that advanced models align with real-world human needs.

This week's book is "Behavioral Network Science: Language, Mind, and Society" by T. T. Hills. You can find all the previous book recommendations on our website. In this week's video, we have a lecture on How We Build Effective Agents by Barry Zhang of Anthropic.

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

"Behavioral Network Science: Language, Mind, and Society" by T. T. Hills successfully bridges two distinct scientific domains, demonstrating how network analysis can reveal hidden patterns in human behavior. The book tackles an impressive scope of topics, from language evolution and childhood learning to cognitive aging, creativity, and social dynamics, while maintaining remarkable coherence throughout. What sets this work apart is Hills' commitment to practical application, equipping readers with concrete tools, including an introductory guide to network science and accompanying R code that enables hands-on analysis.

This practical approach makes the book uniquely valuable to a diverse audience. Behavioral scientists unfamiliar with network methods will find an accessible entry point, while data scientists can discover rich applications in behavioral research. Hills demonstrates particular skill in addressing contemporary social issues through a network lens, offering fresh perspectives on polarization, echo chambers, and conspiracy theories. The interdisciplinary framework proves especially powerful when examining how individual cognitive processes scale up to shape collective behavior and social structures.

The book's most significant achievement lies in its clarity without oversimplification. Hills effectively conveys complex concepts with precision while maintaining an engaging and accessible tone. This balance makes "Behavioral Network Science" essential reading for anyone seeking to understand how network structures influence human behavior across scales—from individual minds to entire societies.

1. ​Agentic Coding Recommendations [lucumr.pocoo.org]
2. ​Design Patterns for Securing LLM Agents against Prompt Injections [simonwillison.net]
3. ​Writing documentation for AI: best practices​ [docs.kapa.ai]
4. ​Introduction to the A* Algorithm​ [redblobgames.com]
5. ​Time Series Forecasting with Graph Transformers​ [kumo.ai]
6. ​Building effective agents​ [anthropic.com]
7. ​The Gentle Singularity [samaltman.com]

• ​Cognitive representations of social networks in isolated villages​ (E. Feltham, L. Forastiere, N. A. Christakis)
• ​ScITree: Scalable Bayesian inference of transmission tree from epidemiological and genomic data​ (H. Waddel, K. Koelle, M. S. Y. Lau)
• ​Clinical knowledge in LLMs does not translate to human interactions​ (A. M. Bean, R. Payne, G. Parsons, H. R. Kirk, J. Ciro, R. Mosquera, S. H. Monsalve, A. S. Ekanayaka, L. Tarassenko, L. Rocher, A. Mahdi)
• ​Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce​ (Y. Shao, H. Zope, Y. Jiang, J. Pei, D. Nguyen, E. Brynjolfsson, D. Yang)
• ​Scaling On-Device GPU Inference for Large Generative Models​ (J. Tang, R. Sarokin, E. Ignasheva, G. Jensen, L. Chen, J. Lee, A. Kulik, M. Grundmann)
• ​Reasoning Language Models: A Blueprint​ (M. Besta, J. Barth, E. Schreiber, A. Kubicek, A. Catarino, R. Gerstenberger, P. Nyczyk, P. Iff, Y. Li, S. Houliston, T. Sternal, M. Copik, G. Kwaśniewski, J. Müller, Ł. Flis, H. Eberhard, Z. Chen, H. Niewiadomski, T. Hoefler)
• ​Unsupervised Elicitation of Language Models​ (J. Wen, Z. Ankner, A. Somani, P. Hase, S. Marks, J. Goldman-Wetzler, L. Petrini, H. Sleight, C. Burns, H. He, S. Feng, E. Perez, J. Leike)

How We Build Effective Agents

​

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

​Unsubscribe​