π¨π° π³ππ πͺ π£πΎπΌ π€π’π€π₯
And the winner isβ¦
Dan Meyer wins the π¨π° π³ππ ππ°ππ«π for Best Take on ChatGPT's Birthday
Illustration is Hard, Especially About the Future
Long-time readers of π¨π° π³ππ know I think illustrating our AI future is as important as writing it. Is My Toddler a Stochastic Parrot? by Angie Wang the best thing I've read about AI since Ted Chiang's Will A.I. Become the New McKinsey? Both were published in the New Yorker.
π¨π° π³ππ ππ‘π’π§π€π¬
AI detection is a confidence game run by companies whose business model is in trouble. I revisit the problems with the way plagiarism detection sets up teachers using AI detectors to assume students have cheated.
Ethan Mollick posts
Instead of developing technology to monitor and control student writing, I hope we follow this advice for teaching as LLMs get better.
The risks of using LLMs to organize your data or your institution's data are significant and not well understood.
Environmental impact matters
The carbon emissions costs of using large models are significant and difficult to measure because giant tech companies will not share information. Melissa HeikkilΓ€ writes about Sasha Luccioni's work on estimating these costs on the open models at Hugging Face. While Iβm plugging HeikkilΓ€, here is a piece she co-authored about Googleβs plans to release Gemini incrementally.
Research notes
Lance Eaton inaugurates Research Insights, where he tries out a lit review format for those of us who want to keep up with the latest on higher ed and generative AI.
Fridays are for puppies
Fridays are for jokes about puppies teaching lessons about AI but really, Fridays are for puppies.
