Researchers · Interviewed at COLM 2026

Chantal Shaib

MIT

COLM 2026 in SF | LLM Judges Can’t Spot AI Slop. Humans Can.

COLM 2026 · October 10, 2026 · 11 min

COLM 2026 in SF | LLM Judges Can’t Spot AI Slop. Humans Can.

Everyone talks about "AI slop," but can you actually measure it? 🤔

In this WiAIR episode recorded live at COLM, Jekaterina Novikova talks with Chantal Shaib, a postdoc at MIT working on AI and writing, about her poster on measuring AI slop in text. The Oxford Dictionary describes slop as low-quality, inaccurate, or unwanted content. But what does "unwanted" look like as a text feature you can actually compute?

More from COLM 2026

All event interviews →