← Back to research

Research Brief · 2026-10-20 Adjacent

DIVE: Benchmarking Full-Duplex Interaction with Visual Engagement in Continuous Streams

The idea, in brief

Its proactive-intervention subset tests useful speech triggered by visual or semantic cues. Timing, participation and content are separate; other duplex tasks are not all proactive.

Editorial summary · not the author abstract. Read the linked paper for the complete abstract and claims.
DIVE: evaluating interaction decisions in duplex conversations
Figure 4 — DIVE: evaluating interaction decisions in duplex conversations. Paper Figure 4 · authors ↗