Demat ! 🇫🇷

In a short but insightful blog post, Katie Bauer, Data Science Manager at Twitter (see ⚡️Trendbreak #1⚡️), draws an interesting parallel between the role of a data analyst and an explorer venturing into uncharted territories. If you're looking for smart takes on data-related issues, definitely give her Twitter account a follow.

Ever played around with image generation using machine learning? If not, let me introduce you to Artbreeder. It's an intriguing platform that lets you create images using a "prompt" (a description of the image content in English) and a "collage" (a very simple arrangement of the main shapes). Here are a couple of my recent masterpieces:
"A lobster doing a zoom call with his boss on his computer, on the beach, in front of the crashing waves" "A data robot eating butter cookies in front of bright sun rays, soviet propaganda style"

For several years now, the scientific community has been grappling with the infamous "reproducibility crisis". Many fields, from sociology to medical sciences to behavioral psychology, have struggled to independently replicate key research findings, leading to a wave of skepticism even for studies published in prestigious journals.
Interestingly, it's not just traditional statistical methods under fire. A recent Twitter thread by Princeton professor Arvind Narayanan and his PhD student, Sayash Kapoor, highlights the potential misuse of machine learning techniques across disciplines, including the prediction of civil wars based on socio-economic indicators in political science. They've put together a workshop on this topic scheduled for July 28. You can check out a summary of their key findings here, and for the more academically inclined, their pre-publication paper is worth a read.

Enjoy the reads, and have a fantastic week! 📚🏝