​(view in browser)​

Dear Reader,

Welcome to the August 8th edition of the Data Science Briefing!

This week’s newsletter features game-changing advances and fresh debates from the heart of AI research. Anthropic’s new work on persona vectors offers precise ways to monitor and control the character traits of large language models, helping developers intervene before harmful personality shifts occur and enabling proactive alignment to human values.

Meanwhile, Google’s MLE-STAR agent sets a new bar for machine learning automation, outperforming rivals by orchestrating web search with fine-grained code refinement to tackle diverse ML tasks and win medals in 63% of Kaggle competitions. DeepMind unveils AlphaEarth Foundations, an AI-powered virtual satellite that integrates petabytes of Earth observation data, mapping land and sea changes across the globe on demand and providing an unmatched resource for climate, agriculture, and development analysis.

On the data front, researchers warn that a major AI training set contains millions of images with sensitive personal details, reigniting privacy concerns over web-scraped data and the urgent need for responsible dataset curation. Also, organizations are rapidly adopting Google’s Gemini Embedding model, harnessing context engineering and retrieval-augmented generation to achieve notable gains in accuracy, efficiency, and multilingual support for enterprise AI applications.

This week’s academic round-up spans insightful breakthroughs across psychology, neurology, epidemiology, and artificial intelligence. The psychophysics of style recasts style perception as an active parsing of form from content, revealing new psychophysical phenomena that shape how we experience and categorize visual aesthetics. In global aging research, Fjell et al. overturn a long-held tenet by demonstrating that higher levels of formal education do not slow cognitive decline or brain aging, while education confers initial advantages in memory and brain structure, its effects do not buffer the inevitable trajectory of age-related decline, underscoring the nuanced role of early-life factors in lifelong cognitive health.

Meanwhile, we explore how multi-group interactions can make epidemic spread contingent on exceeding a threshold of initial infections, offering new insight into the complexity of social contagion. A survey of context engineering for large language models highlights how advances in context retrieval, management, and memory systems pave the way for more adaptive AI, yet also identifies a key limitation. While models now understand context with great nuance, they struggle to generate equally sophisticated outputs, indicating fertile ground for future research.

​Human mobility emerges as central in epidemic modeling, as Lu and co-authors synthesize cutting-edge data-driven methodologies that integrate real-world movement into predictions, leading to transformative improvements in risk assessment, contact tracing, and public health strategy. Finally, “Your Brain on ChatGPT” warns that habitual outsourcing of cognitive work to AI assistants can foster cognitive debt, where users gain fast results but suffer diminished neural engagement, memory, and creativity, reframing the convenience of AI as potentially costly over the long term.

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 by Ambroise Odonnat on Large Language Models as Markov Chains!

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

Michael Lanham's book, "AI Agents in Action", is a practical guide for developers who want to build autonomous AI agents using large language models (LLMs) and open-source frameworks. The book focuses on real-world engineering rather than abstract theory, offering a step-by-step approach to building agent architectures, managing multi-agent systems, and using LLMs to solve business problems. It's written for developers and technical professionals who have the necessary foundational skills in Python and want to move from theoretical knowledge to hands-on development.

The book's strength lies in its gradual layering of complexity, starting with basic concepts and moving to advanced topics like multi-agent orchestration and prompt engineering. Lanham uses open-source tools like CrewAI, AutoGen, and Nexus, and includes annotated code examples to help readers follow along. This approach effectively bridges the gap between academic theory and practical development, making it a valuable toolkit for machine learning engineers who want to create production-ready solutions for tasks like workflow automation and customer service bots. The book also provides insightful commentary on integrating key components like memory and feedback loops into agent-based systems.

However, the book has some notable limitations. A major critique is its optimistic portrayal of the tools and techniques, often overlooking critical discussions about their limitations, trade-offs, and performance at scale. It focuses on illustrative projects rather than addressing issues of robustness and reliability, which are crucial for high-stakes, enterprise-grade deployments. Another drawback is the lack of extended use cases or full-scale system integration examples, which would provide a more complete understanding of an agent system's lifecycle, maintenance, and long-term performance in a real-world business environment.

1. ​Persona vectors: Monitoring and controlling character traits in language models [www.anthropic.com]
2. ​MLE-STAR: A state-of-the-art machine learning engineering agent [research.google]
3. ​AlphaEarth Foundations helps map our planet in unprecedented detail [deepmind.google]
4. ​google/langextract: A Python library for extracting structured information from unstructured text using LLMs with precise source grounding and interactive visualization. [github.com]
5. ​PyTorch in One Hour: From Tensors to Training Neural Networks on Multiple GPUs [sebastianraschka.com]
6. ​When ChatGPT Broke an Entire Field: An Oral History [www.quantamagazine.org]
7. ​A major AI training data set contains millions of examples of personal data [www.technologyreview.com]
8. ​Gemini Embedding: Powering RAG and context engineering [developers.googleblog.com]

• ​The psychophysics of style​ (T. Boger, C. Firestone)
• ​Reevaluating the role of education on cognitive decline and brain aging in longitudinal cohorts across 33 Western countries​ (A. M. Fjell, O. Rogeberg, Ø. Sørensen, I. K. Amlien, D. Bartrés-Faz, A. M. Brandmaier, G. Cattaneo, S. Düzel, H. Grydeland, R. N. Henson, S. Kühn, U. Lindenberger, T. H. Lyngstad, A. M. Mowinckel, L. Nyberg, A. Pascual-Leone, C. Solé-Padullés, M. H. Sneve, J. Solana, M. Strømstad, L. O. Watne, K. B. Walhovd, D. Vidal-Piñeiro)
• ​Phase transition in the susceptible-infected model on hypernetworks​ (G. Fibich, G. Rothmann)
• ​A Survey of Context Engineering for Large Language Models (L. Mei, J. Yao, Y. Ge, Y. Wang, B. Bi, Y. Cai, J. Liu, M. Li, Z.-Z. Li, D. Zhang, C. Zhou, J. Mao, T. Xia, J. Guo, S. Liu)
• ​AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label data (C. F. Brown, M. R. Kazmierski, V. J. Pasquarella, W. J. Rucklidge, M. Samsikova, C. Zhang, E. Shelhamer, E. Lahera, O. Wiles, S. Ilyushchenko, N. Gorelick, L. L. Zhang, S. Alj, E. Schechter, S. Askay, O. Guinan, R. Moore, A. Boukouvalas, P. Kohli)
• ​Human Mobility in Epidemic Modeling (X. Lu, J. Feng, S. Lai, P. Holme, S. Liu, Z. Du, X. Yuan, S. Wang, Y. Li, X. Zhang, Y. Bai, X. Duan, W. Mei, H. Yu, S. Tan, F. Liljeros)
• ​Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task (N. Kosmyna, E. Hauptmann, Y. T. Yuan, J. Situ, X.-H. Liao, A. V. Beresnitzky, I. Braunstein, P. Maes)

Large Language Models as Markov Chains

​

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

​Unsubscribe​