Ethan Mollick posits that an Artificial Superintelligence (ASI) could not create an AI watermarking tool that would be undetectable by another ASI. He argues that any such tool would inherently reveal information about the AI that created it, making it vulnerable to detection by a sufficiently advanced intelligence. Therefore, the development of an un-evadable AI watermark is considered impossible. AI
IMPACT Raises questions about the feasibility of AI watermarking against future superintelligent systems.
RANK_REASON Opinion piece by a named credible voice discussing a hypothetical AI capability.
Read on Bluesky Jetstream — AI desk →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
