AIToday
AI Safety & AlignmentLessWrong AIPublished: Aug 24, 2026, 10:00 JST1 min read

Measure problem gains a third option

Measure problem gains a third option

Key takeaway

  • A third option in the "measure problem" has been proposed.

  • It moves beyond the two main approaches.

  • This could shift how we think about possible realities.

3 Key Points

  1. What happened

    A new post on LessWrong suggests a third approach to the "measure problem," which concerns how to weigh possible realities beyond the two established positions.

  2. Why it matters

    The two existing approaches have significant drawbacks: a simplicity prior leads to counterintuitive implications, while a caring measure ties probabilities to preferences, both of which the new option may address.

  3. What to watch

    The post is niche and assumes background knowledge, so its impact on the broader debate may be limited; however, it could influence future discussion on the topic.

Ask the AI about this article →

Context & Analysis

The measure problem debate on LessWrong has long been framed as a choice between two options: an objective simplicity prior, favored by thinkers like Max Tegmark and Jürgen Schmidhuber, and a caring measure, associated with Wei Dai, Paul Christiano, and Scott Garrabrant. Both have notable drawbacks, as the simplicity prior can lead to strange implications and depends on arbitrary choices like the Universal Turing Machine, while the caring measure ties probabilities and existence to preferences. The new post introduces a third option, but its details and implications are not fully explained, and the post itself is niche, acknowledging that it may lack context for some readers. This suggests the proposal is aimed at a specialized audience and may not immediately reshape the debate, but it could offer a fresh perspective that addresses the shortcomings of the existing positions.

FAQ

What are the two existing approaches to the measure problem?
The two approaches are an 'objective' prior (like a simplicity prior) and a 'caring measure' that ties probabilities to preferences.
What are the drawbacks of the existing approaches?
A simplicity prior leads to counterintuitive implications, like caring more about people easier to find, and depends on arbitrary implementation choices. A caring measure may be problematic because it links existence to preferences.

Get the latest AI Safety & Alignment news every morning

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · takes 30 seconds · unsubscribe anytime

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleEnterprise AI agents meet messy document reality