Ethereum co-founder Vitalik Buterin says AI is extremely knowledgeable but not very wise. In his view, chatbots spot references to almost anything, yet they judge only the surface of what a text argues.
His remarks landed in an X debate over whether chatbots now rival the best human readers. That gap matters for crypto traders who feed whitepapers and governance proposals into AI tools and act on the summaries.
A Harvard Class Without Chatbots Sparked the Debate
The thread started with a Wall Street Journal report on Humanities 10, a demanding Harvard course for freshmen. According to excerpts shared on X, applications rose 40% this fall, with 180 students competing for 90 seats.
Students face a heavy book each week, and the course bans devices. Several applicants said they wanted an antidote to AI shortcuts.
In response, a user argued today's models beat any human as discussion partners on classics because they have read everything. However, a writer pushed back, calling chatbot feedback on her own essays "slop" for now.
What Vitalik Buterin Sees Missing in AI Judgment
Buterin then joined the thread and drew a line between knowledge and wisdom.
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.
— vitalik.eth (@VitalikButerin) October 29, 2026
This is particularly true when it's analyzing writing…
AI’s knowledge, he argued, is easier to spot than its limited wisdom, since wisdom, unlike intelligence, may by definition be hard to verify.
The weakness grows, he added, when a model reads a text for the first time. Famous works, by contrast, arrive with plenty of human analysis in the training data.
His caution fits a pattern. Recently, Buterin warned AI could weaken key parts of crypto security within two years. He has also built a paranoid mode for AI chats.
Not everyone shares his skepticism. Some investors argue that AI already beats top human minds in math.
Meanwhile, crypto teams keep handing audits, research, and governance summaries to AI agents. A surface-level read there could miss a subtle flaw in a smart contract or a token's supply schedule.
Buterin's point suggests facts are easy to check against a source, while judgment about a brand-new text is not. His recent backing of an AI model limiting hacking shows the same instinct to keep human checks in the loop.
Read the Original story Vitalik Buterin Says AI Is Knowledgeable but Not Very Wise by Phil Haunhorst at beincrypto.com