Fact check
“AlphaGo’s policy network was trained on human expert moves.”
Reasoning
The excerpt from MIT Technology Review describes AlphaGo's architecture, noting that the policy network was trained to predict strong human moves. This aligns precisely with the claim that the policy network was trained on human expert moves.
On confidence: The provided source explicitly states that AlphaGo's policy network was trained to guess moves a strong human would play, directly confirming the claim.
Important context
The policy network serves as the 'intuitive' component of AlphaGo, learning from human expert game data to predict plausible moves before deeper search is applied.
Evidence
Supporting (1)
- Tier 1 — Primary sourceDon’t be fooled—LLMs don’t reason | MIT Technology Review
capabilities of this kind. AlphaGo is made up of two systems. The first, its policy network, was trained to guess what move a strong human would play. This “intuitive” part regarded move 37 as nothing special—a play that had a roughly one…
Limitations
The evidence is limited to a single secondary source; no primary research paper is cited. However, the statement is clear and unambiguous regarding the training methodology. No independent origin corroborated this claim, so it cannot be verified and confidence cannot be high.
- Last verified:
- Oct 2, 2026, 4:06 AM CDT
- Pipeline:
- 2.1.0
- Claim type:
- Factual
Where this claim appeared
Don’t be fooled—LLMs don’t reasonMIT Technology Review · Thore Graepel