1 min read
Add as a preferred source on Google

Buterin Says AI Makes Surface-Level Judgments About Writing

He named no AI model or test results. A separate July authorship exercise assessed whether AI could identify who wrote a text.

Mentioned assets
A magnifying glass rests over layered, unmarked pages. / TokenPost.ai
A magnifying glass rests over layered, unmarked pages. / TokenPost.ai

Vitalik Buterin, Ethereum’s co-founder, said Oct. 29 that artificial intelligence (AI) can spot references across subjects but may offer only a shallow reading of an unfamiliar text’s point.

“I find AI to be extremely knowledgeable, and so exceptionally able to detect references to almost anything, but not very wise, and so it only makes the most surface-level judgements about what point some writing is making,” Buterin said.

His post named no AI model or test results. It also gave no measured comparison with human readers.

A separate July authorship exercise tested a different task. Franklyn Wang said an AI-assisted analysis assessed 27 texts and gave Buterin roughly 20% confidence as the author of an anonymous rewrite of Ethereum Improvement Proposal 7503 (EIP-7503). Buterin confirmed the identification, saying the model recognized his habits in explaining mathematics and algorithms despite his effort to disguise his prose.

The July exercise tried to identify who wrote a text. Buterin’s Oct. 29 comment addressed how AI judges what writing argues, so the authorship result does not establish how well AI evaluates an unfamiliar text’s argument.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…