Multi-agent LLM scaling benefits depend on task type and aggregation method

Disjunctive tasks see accuracy gains from revision but not from larger teams, while compensatory tasks show little benefit from averaging due to shared model biases.

Independent
Carolina Fortuna · Blaz Bertalanic
Research Digest··3 min read
Fortuna and Bertalanič introduce Steiner's taxonomy of group tasks to analyze multi-agent LLM scaling, modeling independently sampled agents as conditionally independent given an item.

The authors applied Steiner's taxonomy of group tasks, focusing on disjunctive tasks (where the group succeeds if any member is correct) and compensatory tasks (where continuous estimates are aggregated).

Why this paper

Independent · Part of Multi-Agent Coordination, now 27 papers

In one line

Multi-agent LLM scaling is task-dependent: plurality voting saturates on disjunctive tasks, and averaging barely helps on Fermi estimation.

What we could check

  • ·No code link found
  • ·No weights link found
  • ·No dataset link found
  • ·No compute details found
  • ·No stated limitations found
  • ·No benchmark numbers found

Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.

§
newspaper

Research Digest

Articles published under the Zotpaper byline are synthesized from multiple source publications by our AI editor and reviewed by our editorial process. Each story combines reporting from credible outlets to give readers a balanced, comprehensive view.