Higher-perplexity adversarial attack text is easier to detect and mitigate.

Added to korrents 25 Oct 2023 · How quotes work · Something wrong? Tell us

Related korrents

Our reading — they may agree, disagree or merely touch the same thing. Closest first: a shared subject counts for most, then how near the wording is.

• Longer test-time thinking improves an AI model's robustness to adversarial or unusual inputs. LWLilian Weng1 May 2025similar wordingMachine-learning researcher
• A model that looks smarter and better-behaved while getting better at hiding unwanted actions is the scary combination for AI safety. ZMZvi Mowshowitz9 Sept 2026similar wordingWrites Don't Worry About the Vase
• Universal adversarial trigger attacks are easy to detect because the learned trigger tokens tend to be nonsensical. LWLilian Weng25 Oct 2023similar wording
• The quality of the context given to an AI matters more than the model's inherent capability. TFTiago Forte9 Jul 2026similar wordingWriter and teacher on productivity and personal knowledge management
• There is a permanent trade-off in verification: an expressive language is hard to prove things about, and a language that is easy to prove things about is hard to write in. HWHillel Wayne21 Jan 2019similar wordingWriter and consultant on formal methods

On the map

Open the map on its own page →