📢

Do mesa-optimizer risk arguments rely on the train-test paradigm?

📅 2020
public awareness
🔵 Rare
#ai #mesa-optimization

📖 Description

Going by the [Risks from Learned Optimization sequence](https://www.alignmentforum.org/s/r9tYkB2a8Fp4DN8yB), it's not clear if mesa-optimization is a big threat if the model continues to be updated throughout deployment. I suspect this has been discussed before (links welcome), but I didn't find anything with a quick search.

📊 Game Impacts

Variable Change Condition
Research +5 Always
Vibey Doom +5 Always
Ethics Risk -5 Always

💭 Reactions

🔬 Safety Researcher Reaction: ⚠️ Placeholder - Needs Real Quote
"Critical insights for the field"
📰 Media Reaction: ⚠️ Placeholder - Needs Real Quote
"Discussed in AI safety community"
💡 Found a Real Quote? Suggest it here

🔗 Sources

🏷️ Event Metadata

Think this event's metadata could be improved? Suggest changes to category, rarity, tags, game impacts, or p(doom) effects.

📁 Category Public Awareness → Suggest different category
⭐ Rarity 🔵 Rare → Suggest different rarity
🏷️ Tags (2) ai, mesa-optimization → Suggest tag changes
📊 Game Impacts (3) 3 variable changes → Suggest impact changes
☢️ p(Doom) Impact No direct impact → Suggest p(doom) change
📝 General Metadata Year, description, reactions → Comprehensive review

🤝 Found an Issue?

This event data is sourced from the pdoom-data repository. If you notice errors or want to suggest improvements:

GitHub Issue (Preferred) 📧 Email (No GitHub)
← Back to All Events