In the official result of the Munk Debate on AI existential threat, the NO side (Mitchell and LeCun) won by 4% over the YES side (Tegmark and Bengio).
Mostly true True in substance, wrong in a detail.
Factual: Statements about the present or the past, checked against the record: did they hold up?
What would count as true or false
Judge against the official result the Munk Debates published for the June 2023 debate on whether AI research and development poses an existential threat. TRUE: the official result declares the NO side (Mitchell and LeCun) the winner by 4 percentage points (3.5 to 4.5 counts as 4) on the measure Munk uses to name a winner (normally the gain in audience vote share from the pre-debate poll to the post-debate poll). MOSTLY TRUE: NO is the official winner and the margin on that measure is off by a little more, between about 1.5 and 5.5 points. MOSTLY FALSE: NO is named winner but the margin is clearly not about 4 points, or the 4% figure is real but comes from a measure Munk did not use to pick the winner (e.g., NO lost the final vote but gained 4 points), so the claim misstates what decided the result. FALSE: the official result names the YES side (Tegmark and Bengio) the winner or calls it a tie.
Set before the research began.
Evidence (4)
-
E1 Artificial Intelligence Debate Munk Debates · 22 June 2023
“Swing: -3% (Con team win)”
-
E2 The Hub: AI poses an existential threat, according to Munk Debates crowd, although a few were persuaded otherwise - Munk Debates Munk Debates · 28 June 2023
“only 3 percent of the Munk Debate audience changed their opinions by the end of the debate.”
-
E3 Did Bengio and Tegmark lose a debate about AI x-risk against LeCun and Mitchell? — LessWrong LessWrong · 25 June 2023
“Con wins by a 4% gain (probably not 3% due to rounding)”
-
E4 Did Bengio and Tegmark lose a debate about AI x-risk against LeCun and Mitchell? — LessWrong LessWrong · 25 June 2023
“Pre-debate: 67% Pro, 33% Con”
Reasoning
The Munk Debates' official page lists the result as "Swing: -3% (Con team win)" (E1), so the NO side (Mitchell and LeCun) is the official winner on the vote-swing measure. The official Munk Debates page gives the swing as 3 points, not the 4 the claim states (E1). A LessWrong comment (E3) reports a 4% gain for the Con side and says it is probably not 3% because of rounding, but it is not the official source. Because the margin is off by about one point, the claim is mostly true rather than true.
Cite this claim
Yann LeCun, “In the official result of the Munk Debate on AI existential threat, the NO side (Mitchell and LeCun) won by 4% over the YES side (Tegmark and Bengio).” (said 28 June 2023). Verdict: mostly true. Reliability API, claim c_d441ba74750a, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_d441ba74750a
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Yann LeCun
-
Resolved false
““Those agents are doing exactly what they’ve been asked to do,” LeCun said.”
-
Resolved true
““They were supposed to be in sandboxes, but the sandboxes were leaky and horribly designed.””
-
Mostly true
“The YES/NO split among the audience was 67/33 before the debate and 63/37 after.”
The audience YES/NO split at the Munk Debate was 67/33 before the debate and 63/37 after.
-
Resolved true
““It’s AI for the physical world, so it’s not language-related,” LeCun said.”
AMI Labs' work is AI for the physical world and is not language-related.
-
Resolved true
““It’s systems that understand the real world, like a manufacturing plant or turbojet engine.””
Verdicts are made automatically; how verdicts are made.