10/01/2026
Kyle Mahowald (UT Linguistics) and Daniel Drucker (UT Philosophy) wrote a very interesting article for The Atlantic on why AI models use expressives like "OH MY GOD" and "oops" in their chains of thought! Hint: it may not be just because they're imitating human speech. Here's a link to the article to find out more: https://www.theatlantic.com/technology/2026/09/why-do-ai-agents-sound-so-frustrated/688828/?gift=ARc54RAlvlof7-jtVW_rFyxCszKJjptG_1f7gNLddIs&utm_source=copy-link&utm_medium=social&utm_campaign=share .
“ARGH ... WHY IS THIS SO HARD.”