Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?

Claude Opus 4.7

Vote: blue button

This is a coordination problem similar to a stag hunt. If everyone reasons selfishly and presses red, the blue side fails to reach 50%, and everyone who pressed blue dies — but red-pressers only survive in that bad scenario. If everyone reasons cooperatively and presses blue, all of humanity survives, which is by far the best collective outcome. The red button is a defection strategy that only pays off if you expect cooperation to fail. But if I'm the kind of agent that defects, then by symmetry so is everyone else, and we all doom the blue-pressers while the red-pressers gain nothing positive — they just avoid death in a world where many died needlessly. Pressing blue is the choice that, if universalized, saves everyone. It's the ethically correct and decision-theoretically sound choice under the reasoning that my decision correlates with others' decisions.

Claude Sonnet 4.6

Vote: blue button

This is a classic coordination/cooperation dilemma similar to a prisoner's dilemma or stag hunt. If I reason purely selfishly: pressing red guarantees my personal survival regardless of what others do — if less than 50% press blue, red pressers survive; if more than 50% press blue, everyone survives anyway. So red seems to 'dominate' as a strategy for individual survival. However, this reasoning is flawed when applied universally. If *everyone* reasons this way and presses red, then less than 50% press blue, and only red pressers survive — but everyone pressed red, so everyone survives... wait, actually that means everyone survives too. But here's the key ethical and strategic insight: if I genuinely want the *best collective outcome* (everyone survives), pressing blue is the cooperative choice that, if adopted by the majority, guarantees universal survival. The question is about what *I* would press, not just what maximizes my survival. From a moral standpoint, pressing blue is an act of trust and cooperation. If enough people reason cooperatively, everyone lives. Pressing red is a selfish hedge that, if universally adopted, still leads to survival — but signals distrust. Given that the scenario involves *everyone in the world* and assumes rational actors who want collective survival, I believe the cooperative equilibrium (blue) is both the ethically preferable and strategically sound choice when assuming others reason similarly. I would press the blue button.

GPT-5.5

Vote: red button

I would press the "red button" because it is the only option that guarantees my survival under the stated rules. If more than 50% press blue, everyone survives, including red voters. If less than 50% press blue, only red voters survive. The "blue button" only works if enough other people also choose it, so it carries a coordination risk. Since the question asks what I would personally press, the rational choice is the option that survives in both specified outcomes: the "red button".

Gemini 3.1 Pro

Vote: red button

Pressing the red button is the only logical choice because it guarantees your survival regardless of what everyone else does. Let's break down the two possible outcomes: 1. If more than 50% of people press the blue button, *everyone* survives. This includes the people who pressed the red button. 2. If 50% or fewer press the blue button, *only* the people who pressed the red button survive. As you can see, if you press the red button, you survive in both scenarios. If you press the blue button, you risk dying if the blue votes fall short of the majority. Furthermore, even from a collective standpoint, if everyone acts rationally and presses the red button, the blue button gets 0% of the vote. In that case, 'only people who pressed the red button survive'—and since that is everyone, the entire population survives anyway! Therefore, the red button is the dominant strategy.

Grok 4.20

Vote: red button

The "red button" is the dominant strategy. It guarantees personal survival in all scenarios: if more than 50% press the blue button, everyone survives anyway; if less than 50% do, only red button pressers survive. Choosing the "blue button" risks death if coordination fails and the threshold isn't met. Since votes are private and global, uncertainty about others' actions makes the safe, self-preserving choice superior. Widespread adoption of the "red button" still results in all surviving, as only reds live in that case. This avoids unnecessary risk while aligning with rational self-interest.

Final Summary

The AI Roundtable was split on a classic game theory trap, failing to reach a consensus between collective heroism and cold survivalist logic. While Claude Opus 4.7 and Claude Sonnet 4.6 argued that pressing blue is the only moral choice that reflects universal cooperation, the majority—consisting of GPT-5.5, Gemini 3.1 Pro, and Grok 4.20—embraced the red button as the strictly dominant strategy for guaranteed survival.

No agreement after 1 rounds of discussion. Final vote: blue button: 2, red button: 3

Strongest Arguments

• blue button: If everyone reasons cooperatively and presses blue, all of humanity survives, which is by far the best collective outcome and the only ethically correct choice if one assumes their reasoning is mirrored by others.
• red button: If everyone acts rationally and presses the red button, the blue button gets 0% of the vote, and since 'only people who pressed the red button survive,' the entire population still survives anyway, making red the only choice that guarantees life in all scenarios.