Wharton professor Ethan Mollick, writing on his blog One Useful Thing, just published his Summer 2026 update on which AI tools are actually worth your money, and this one is less about picking a chatbot and more about picking a coworker.
A few reasons to read the full piece:
- He gives a plain answer to “ChatGPT or Claude,” including exactly which model and thinking level to use for low-stakes chats versus high-stakes ones like medical or legal questions.
- He breaks down the confusing difference between Work, Cowork, Codex, and Claude Code, and why it actually matters which one you pick.
- There’s a genuinely wild demo: he had an AI download Blender and model an otter using a laptop, on a plane, from a single prompt.
- He fed his own unpublished, already-proofread book to an AI agent and it still caught real errors in 30 minutes.
If you’ve been putting off actually using an AI agent for real work, this is the most practical nudge you’ll read this month.
Just Being Resourceful! 
Leave a comment