We conducted a study on 3691 people to empirically test 40 claims about IQ (including claims made in academia and by the general public)! Here are some of our most important findings (see the link at the bottom of this thread for all 40 results): 🧵

Only one European (well German of course) car in the world's best selling cars now Not a single European company left in the leading tech companies anymore Not a single European company left in the top AI companies This is why Europe has zero negotiation power at the world

Contrary to Machine Learning, it feels like AI is discovered rather than designed.

How has AI coding led to productivity gains? How should we think about code hygiene? Do we still need to review code? Or is it okay to just update code solely via an LLM mediator? Context: I've been writing and reviewing a lot of (Python) code. AI code has a distinct smell. It's

Over 200 people joined our AI-Powered Decision Making under Uncertainty workshop last week. The discord chat and q&a was on fire. Thank you Allen Downey, Chris Fonnesbeck, & PyMC Labs for an enlightening session. Thomas Wiecki got absolutely wild in the chat: > Bayesian modeling,

Allen Downey PyMC Labs Thomas Wiecki livestream here: youtube.com/live/2Auc57lxg… repo here: github.com/pymc-labs/ai_d…

Can LLMs make business decisions you'd actually trust? Thomas Wiecki joins Hugo Bowne-Anderson to share how #PyMC Labs is building GenAI systems for clients like Colgate 🧪 Real-world use cases 📊 90% accuracy 🔁 Closed-loop systems 🔗 dub.link/genai-decision #GenAI #LLMsystems

I had access to GPT-5. I think it is a very big deal as it is very smart & just does stuff for you Full write up in comments, but this is “make a procedural brutalist building creator where i can drag and edit buildings in cool ways" & "make it better" a bunch. I touched no code

GPT-5 is here. Rolling out to everyone starting today. openai.com/gpt-5/

I'm testing GPT-5 with an impromptu self-therapy session, it is *very* good for that. Extremely insightful and concise - every word is placed perfectly.

From cool demos to boardroom decisions — AI is now running analyses, critiquing itself & guiding high-stakes business calls. 🎥 Aug 15, 11 AM EDT — Thomas Wiecki joins Hugo Bowne-Anderson to show how to build AI you can trust. Watch: dub.link/genai-decision #AgenticAI #GenAI

Come hang out and discuss AI, Bayes, how they can elevate each other, and maybe even meditation if we get wild. Starting 5 mins.

Last week I spoke with Thomas Wiecki (PyMC Labs) about "Building GenAI Systems That Make Business Decisions". My mind's still spinning. Watch the livestream here: youtube.com/live/so4AzEbgS… Podcast out soon. Talking points below 👇

meditation might have the worst pedagogy of any field. they solved the cause* and cure** to suffering but hid it behind weird poems and deities * clenching and resistance ** train to not resist, see how you gradually feel way better, see the non-specialness of self-sensations

Most budget planning is just expensive guessing 🎯 Carlos tackles this in "Risk Budget Optimization for Causal Mix Models" @PyDataBerlin 2025! 📅 Sept 1st, 5:00-5:30 PM Turn uncertainty into strategic advantage. #PyDataBerlin #BudgetOptimization #PyMC

Stop waiting weeks for modeling answers! Our new Expert Access Program (EAP) gives your team direct access to our #Bayesian and #PyMC experts for timely answers, guidance, and confidence to move faster. Learn more👉 dub.link/BAbeiyf

Former CEO of Google on how AI is still "underhyped"

why does this happen? the model believes there's a seahorse emoji, sure, but why does that make it output a *different* emoji? here's a clue from everyone's favorite underrated interpretability tool, logit lens! in logit lens, we use the model's lm_head in a weird way.

It’s still a wild idea that you can just give an AI agent a task and leave your computer for a bit and come back and it’s done. The ability to kick off coding, building a prototype, doing deep research, doing data analysis, generating a pitch, or any other common knowledge work