Issue #246

May 29, 2024

​(view in browser)​

​

Dear Reader,

Welcome to the 246th edition of the Data Science Briefing!

The latest medium post on the Epidemiology series: Epidemiology 303: Metapopulation models, where we explore how to connect multiple populations through a travel matrix is now out. You can catch up on the latest in the Graphs for Data Science substack: k-core Decomposition, or in the Viz4Sci series, The Effects of Vaccination.

We're happy to remind you that the very first edition of the LangChain for Generative AI Pipelines is coming up in just a week. Only a few spots are left, so Register now so you don't miss out!

In our regularly scheduled content, we have reflections from a Year of Building with LLMs, detailed introductions to Introduction to Reinforcement Learning, Diffusion Models, and Hypothesis Testing.

On the academic front, we explore how Novel embeddings improve the prediction of risk perception, ponder if artificial consciousness is achievable and learn how to Transform Classification Metrics into Loss Functions.

This week's book recommendation is "Co-Intelligence" by Ethan Mollick. You can find all the previous book recommendations on our website. In this week's video, we have a tutorial on the strange things you see when you Map LLMs like GPT.

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

This week's book is "Co-Intelligence" by Ethan Mollick. This book provides an essential and balanced guide to navigating the age of artificial intelligence (AI). Author Ethan Mollick offers a pragmatic perspective on AI's capabilities and limitations, showing how it can effectively augment human abilities. The book's key strength is Mollick's "Four Rules of Co-Intelligence" framework for seamlessly integrating AI into work and life. He demystifies complex AI concepts through engaging examples and practical advice. Mollick paints an optimistic yet grounded vision where humans and AI collaborate harmoniously, complementing each other's strengths to drive innovation. His book equips readers to confidently leverage AI's power while preserving human ingenuity and ethics. In the rapidly changing AI landscape, "Co-Intelligence" is an invaluable resource for business leaders, educators, students, and anyone seeking to thrive by harnessing the benefits of human-AI co-intelligence. Mollick's work provides a roadmap for gaining a competitive edge through co-intelligent collaboration.

1. ​What We Learned from a Year of Building with LLMs (Part I) [oreilly.com]
2. ​An Introduction to Reinforcement Learning​ [towardsdatascience.com]
3. ​Computer Scientists Invent an Efficient New Way to Count​ [quantamagazine.org]
4. ​Big Data Is Dead​ [motherduck.com]
5. ​Welcome to Microsoft Phi-3 Cookbook​ [github.com/microsoft]
6. ​Diffusion Models​ [andrewkchan.dev]
7. ​Hypothesis Testing Explained​ [towardsdatascience.com]
8. ​These crows may count in a way similar to human toddlers​ [science.org]

• ​Novel embeddings improve the prediction of risk perception (Z. Hussain, R. Mata, D. U. Wulff)
• ​Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet​ (A. Templeton, T. Conerly, J. Marcus, J. Lindsey, T. Bricken, B. Chen, A. Pearce, C. Citro, E. Ameisen, A. Jones, H. Cunningham, N. L. Turner, C. McDougall, M. MacDiarmid, A. Tamkin, E. Durmus, T. Hume, F. Mosconi, C. D. Freeman, T. R. Sumers, E. Rees, J. Batson, A. Jermyn, S. Carter, C. Olah, T. Henighan)
• ​Link Prediction in Complex Networks Using Average Centrality-Based Similarity Score​ (Y. V. Nandini, T. J. Lakshmi, M. K. Enduri, H. Sharma)
• ​A generative model for community types in directed networks​ (C. X. Liu, T. J. Alexander, E. G. Altmann)
• ​Is artificial consciousness achievable? Lessons from the human brain​ (M. Farisco, K. Evers, J.-P. Changeux)
• ​Thermodynamic Natural Gradient Descent​ (K. Donatella, S. Duffield, M. Aifer, D. Melanson, G. Crooks, P. J. Coles)
• ​AnyLoss: Transforming Classification Metrics into Loss Functions​ (D. Han, N. Moniz, N. V. Chawla)

Mapping GPT revealed something strange...

​

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