Frame for Take-Off Speeds to inform compute governance & scaling alignment

📖 Description

Figure 1: Something happens at future time t' that causes more resources to be poured into alignmentThe argument goes: there will be a time in the future, t', where e.g. a terrible AI accident occurs, alignment failures are documented (e.g. partial deception), or the majority of GDP is AI such that more people are pouring resources into aligning AI. Potentially to the point that >90% of alignment resources will be used in the years before x-catastrophe or a pivotal act (Figure 2)

📊 Game Impacts

💭 Reactions

🔗 Sources

• [1] https://www.alignmentforum.org/posts/XFL3vaA69mHxATWM7/frame-for-take-off-speeds-to-inform-compute-governance-and