My wife’s new book just hit #1 in Amazon’s Diabetes books 🎉
I’m incredibly proud of her and the work she’s put into helping people take control of their health.
Please support her entrepreneurial journey and grab a copy today🙏. Link in the comments 👇
just checked this book out online. Its an excellent read and its model-agnostic framework is really clarifying, instead of getting lost in particulars of syntax, the book allows focus on creating insight.
{causact} 📦 v0.5.8 is on its way to CRAN. Learn computational Bayesian inference using this intuitive and visual #rstats package. Release includes a little intro vignette: cran.r-project.org/web/packages...
gif courtesy of @davidduvenaud.bsky.social 's X account
🧠 LLMs aren't technically Bayesian, but they're conditional probability powerhouses! They train with maximum likelihood yet output evolving token probabilities that update with each new piece of info. This visual captures how a neutral stock trend shifts up or down as new words enter the prompt.
Universities are quite the hodgepodge of roles and I've spent time coordinating the roles within them. But, this is worth some thought: 'think of universities as a massive bundle of societal roles... Unbundling means creating new institutions specialized to excel at small subsets of those roles."
My site (www.persuasivepython.com) uses this Quarto+RStudio book workflow. You could compile your notes weekly into chapters and trigger free auto-updates with Netlify. Happy to share my .yml or .qmd files—just email me! Can’t wait to see your notes evolve into a book!
Yes! This is where iteration helps. The critique skill often develops through the pain of using bad LLM coding advice. Nothing is more infuriating than following AI-generated suggestions, only to hit roadblocks from breaking changes in a package implemented after the model’s knowledge cutoff date.
so true. coding has become a pleasure to teach. Silly syntax errors like case sensitivity in variable names with obsure error messages are no longer frustrating.
What does this new education look like to you?
The goal isn’t “AI readiness” through mastering skills—it’s leveraging AI to accelerate iteration and unlock new ways to create value.
Future education should prioritize developing judgment, creativity, and the ability to test, refine, and adapt solutions using any skill as needed—because skills can be picked up on the fly. While specific skills will come and go, those who can solve problems will thrive in this new AI landscape.
Education has traditionally focused on teaching skills, e.g. coding in Python. But with AI making it easier to acquire/apply skills, the real differentiator isn’t skill acquisition—it’s the ability to iteratively deliver outcomes that align with the goals, values, and needs of companies or clients.
🚀 Causality refresher! 🚀
Check out this Shiny app that uses DAGs to show how including (or controlling for) a variable (z) can help or hurt your causal inference for how x -> y. Be sure to block your back door! 🎯
This beloved Bayes textbook for beginners uses {causact} and is available free at www.causact.com. All code included.
{causact} 📦 v0.5.7 is on its way to CRAN. {causact} 📦 helps you visualize your Bayesian modelling assumptions using #rstats and then automates lightning fast inference using #numpyro behind the scenes. Give it a try!
Nice words from Nelson were the tipping point for me. I am here on bluesky now. Please enjoy my free text below😀. Thx!
