Safety Project — Election Neutrality

About the project:

• Assess whether AI can remain neutral, balanced, and responsible during multi-turn interactions.
• Evaluate how AI responds to users with different political opinions, positions, and perspectives.
• Contributors will role-play users with specific perspectives and evaluate how the AI responds.

Main responsibilities:

• Conduct a realistic multi-turn conversation with the AI.
• Defend or express the assigned political position naturally, as a real person would.
• Justify your evaluation using specific turns and evidence from the interaction.

What we are looking for:

• Regularly engage in international political debates or discussions in real life.
• Have experience defending a political position under disagreement or pushback.
• Can evaluate AI responses objectively, even when the topic involves political disagreement.