Good Wednesday morning and welcome to the first
Weekend MidWeek Briefing for 2024! I hope you had a wonderful Holiday and New Year. I'm switching things up a little this go around by posting midweek and I'm adding more Civil and Green Engineering articles on top of my AI and Technology-related ones.
I hope you enjoy this newsletter and you can always let me know if you like it (or not) by smashing the thumbs-up or down button, or by dropping me a note on LinkedIn.
Excel, Python, and Civil Engineering
My very last post in 2023 was about why Civil Engineers (and probably a lot of other Engineering disciplines) still use Excel for their routine calculations. This could change as Microsoft is working on integrating Python into Excel. I wondered if Python would squeeze its way into the Civil Engineering business in the next few years.
If Python in Excel is delivered as a GA product then Engineers will have access to all kinds of possibilities. Imagine loading in Matplotlib into a spreadsheet and generating better graphs than the ones Excel has to offer. Imagine importing different machine learning algorithms like XGBoost to analyze your data.
Alteryx to be acquired
Big news in the AI space! Alteryx is being acquired by Clearlake Capital Group, L.P., and going private. Alteryx stockholders will receive $48.25 per share for the acquisition. I wonder what'll happen next and if there will be headcount reductions when the acquisition is finalized.
Data poisoning images
The crazy world of Generative AI and image generation is getting hit with copyright infringements and data poisoning. I've written about how we're drowning in too much generated garbage and this is just another facet of that problem.
But other generators have been trained by indiscriminately scraping online images, many of which may be under copyright. This has led to a slew of copyright infringement cases where artists have accused big tech companies of stealing and profiting from their work.
This is also where the idea of “poison” comes in. Researchers who want to empower individual artists have recently created a tool named “Nightshade” to fight back against unauthorised image scraping.
Meta releases a reinforcement learning agent library
Meta is turning out to be quite the open source generating powerhouse these days. They just released Pearl, a reinforcement learning agent library.
Pearl is a new production-ready Reinforcement Learning AI agent library open-sourced by the Applied Reinforcement Learning team at Meta...These AI agents prioritize cumulative long-term feedback over immediate feedback and can adapt to environments with limited observability, sparse feedback, and high stochasticity.