ICYMI StanCon '24 Sept 10-12 at the Oxford School of Witchcraft and Probabilistic Programming, now accepting proposals for contributed talks and tutorials. Hope to see Stan stans there mc-stan.org/events/stancon…

I really enjoyed Cal's recent interview with Tim Ferriss and the below review of key points from Cal's book is great. If, like me, you are at your best when curiously wandering, this is a good reminder on how to protect that wonderful state.

Two comparisons of data analysts to Code Interpreter: "Experimental results show that GPT-4 can achieve comparable performance to humans" arxiv.org/pdf/2305.15038… GPT-4 scores over 90% on exams, the data science field is “on the verge of a paradigm shift” arxiv.org/pdf/2307.02792…

Who’s up for a nice thread about narratively generative modeling?

media.rff.org/documents/RFF-…

gsb.stanford.edu/faculty-resear…

I’ve see this phenomenon a lot. Someone identifies a problem that it seems LLMs cannot solve but humans can, and then it turns out that better prompting is enough to solve it easily This even happened to me. I thought GPT-4 couldn’t do crosswords. Arvind Narayanan proved me wrong

Yes yes confusion is what thinking feels like

Are you a UD grad student or postdoc seeking to hone your communication skills and translate your research for broad audiences? Join us in Warner Hall, Rm 101 on Thurs., 4/25 for a workshop led by Katie Ryan from LanguageConnectED. Space limited. RSVP at bit.ly/3W5iSBs

Ever wonder if Perplexity AI knows about newer #rstats packages? The answer is "yes". Example: do you know anything about the causact package in R? perplexity.ai/search/do-you-…

I am three months late on the news that Neil Young and Joni Mitchell are back on Spotify, but I am a very happy music listener today. cbsnews.com/news/neil-youn… via CBS News

Mathematicians Shooting Pool: Is Innovating Without Practitioner Engagement Possible? A lesson for all of us analysts and academics. youtube.com/clip/UgkxhaIsp…

This is true for any abstraction. DAGs are one level of fidelity for describing a data generating process sitting between a natural language description and formal specification. None of them are sufficient, but each let's you discuss and think about a DGP in new context.

A good reminder that publication is the *beginning* of peer review

Doing research on probabilistic modeling, decision-making under uncertainty, or efficient learning and optimization? Come to NeurIPS to present your work! Submissions due August 29th. gp-seminar-series.github.io/neurips-2024/

This is so good by James Clear

I once wrestled with choosing between R and Python, hindered by the syntax barrier. Now, with LLMs like ChatGPT, Claude, or Copilot, I focus on goals: NodeJS for web backends, React Native for mobile apps, Python for data science, and R for statistical analysis. Let AI handle the

Looking for recent work on the Wisdom of Crowds? Search no longer! The articles in this special issue of American Psychological Association's journal Decision on the wisdom of the crowds contribute to outstanding questions regarding what to aggregate, optimal weighting schemes,