Today’s picks
1. AI models may not be so opaque after all
AI models are often described as black boxes whose inner workings defy inspection, a worry at the heart of the safety debate. This piece surveys recent interpretability work suggesting that model internals may be more legible than assumed. It is worth reading for connecting technical findings to the broader question of whether AI can be made accountable.
2. Meet the original ‘Ghost Busters’—the 17th century scientists who investigated hauntings
This article introduces the 17th-century scientists who investigated reports of hauntings, approaching alleged ghosts with observation, documentation, and skepticism rather than fear. Their cases show how early modern inquiry negotiated between popular belief and emerging scientific methods. It is worth reading for a lively, surprising angle on the history of science and the roots of “ghost hunting.”
3. Cover Story newsletter: The agony of liberal Europe
This cover story newsletter examines the predicament facing liberal Europe and offers an exclusive look at how the cover was designed. It combines the political and social tensions behind the story with insight into the editorial choices that shaped its visual presentation. Readers interested in European politics and the craft of news publishing will find a concise, revealing behind-the-scenes account.
4. Strange triangular craft reported in new U.S. government UFO release
The latest batch of declassified U.S. government files includes eyewitness reports from earlier this year describing a black triangular craft performing maneuvers that witnesses could not explain. The accounts are testimonial and do not establish what the object was. The release is worth reading for anyone following unexplained aerial phenomena, aviation safety, and government transparency, while keeping the limits of the evidence in view.