Turn prediction benchmark: best systems have poor recall and high false positives
alexisgallagher · x · 2026-09-13
Citing a recent benchmark of turn prediction systems: the best existing predictors still miss many genuine turn ends (poor recall) and misidentify pauses as ends (high false positives). In the paper's corpus, turn gaps are negative for much human speech due to overlaps — and no voice system implements this accurately, the author notes.
Related event: Study: voice AI turn-taking still far from human-level(3 posts)→
More from AGI Musings
- Schmidhuber maps 40 years of recursive self-improvement, dismissing claims Google just cracked RSI — burny_tech · 2026-09-14
- The AI panic in a nutshell: elites pushing "safety" to protect their interests — AIandDesign · 2026-09-14
- Dev argues ASI won't kill people — real AI safety issue is distributing abundance — tobowers · 2026-09-14
- 10+ emails in a lease negotiation turned out to be his AI talking to their AI — adamamcbride · 2026-09-14
- Eight years'-caliber math results land in eight weeks: zeta bounds, prime gaps, FLT formalization — Zulfikar_Ramzan · 2026-09-14
- Tyler Cowen's Simple Model of AI-Aided Growth: Intelligence Meets Polanyi Knowledge — sebkrier · 2026-09-14