What percentage of Manifold poll respondents will agree that weak AGI has been achieved at the end of June 2024?
Mini
17
2.8k
resolved Jul 1
Resolved
26-28

On June 23, 2024, a poll will be posted that asks Manifold users whether weak AGI has been achieved as of June 23. The poll will remain open until June 30. The answers will be exactly the same as the following poll, but no specific model will be referenced:

/SteveSokolowski/did-openai-achieve-weak-agi-with-gp

This market will resolve to the following number:

ceiling (100.0 * (number of YES votes) / (number of YES + NO votes))

Responses of "no opinion" will be ignored, so users don't vote just to see the results. The poll will intentionally not define "weak AGI," other than to differentiate it from "strong AGI" that manipulates the physical world, because the purpose is to see what society generally believes.

For reference, if this market had existed for the May poll, it would have resolved to 17 at the time of market creation. What will the number be at the end of June?

Get Ṁ600 play money

🏅 Top traders

#NameTotal profit
1Ṁ988
2Ṁ38
3Ṁ10
4Ṁ5
5Ṁ3
Sort by:

The answer for this month was 27. Last month was 17%, but last month's poll only asked about GPT-4o, and Claude 3.5 Sonnet was released since then, so I think the results are right in line with expected progress on the topic. See you next time!

Big leap forward

I don't think so. I just think that other markets on this topic, like the Metaculus market, aren't getting resolved even though the criteria has long been passed. Additionally, people who don't use the models every day underestimate their capabilities and don't realize how quicky they are improving.

This market will be resolved on June 30. If you want to continue betting, another market is still available:

/SteveSokolowski/will-at-least-20-of-manifold-respon

I'm not getting how this market is below 17.

The May poll is at 17, and the question is weak - it only asks whether one model specifically, at a specific time, achieved AGI. There are comments in that poll from people who voted NO solely because they felt that AGI had already been achieved prior to that model. That doesn't take into account Meta's Chamelon paper today or anything else that comes out over the next 35 days.

"been achieved as of June 23"

Do you mean June 24?

@Daniel_MC No. The poll will open on June 23 and last until June 30. If some advance occurs on June 24, the question will remain whether AGI had been achieved on June 23.

@Daniel_MC I think it's intended to mean June 23rd, not June 2023.

@ML12d1 yeah, got done by the American dates