2026-10-24 / Signal #4
Jack Dorsey’s lab trained an AI that learned to talk by listening only — no text, ever
“It never learned to read. It just listened until it could talk back.”
The Story
Dessa is a full-duplex speech-to-speech model trained from scratch on recorded human conversation audio alone — no transcripts, no pretrained text models. It laughs, sighs, backchannels, and interrupts in real time; Block claims it runs locally on a high-end Mac. Early access is limited to safety researchers.
Why It Matters
Identity and language: an intelligence that never read a word, only heard people. Builder-curious (tiny lab, one rack of GPUs) plus future-shock of voice AIs that feel like they grew up in a room instead of a corpus.
Evidence
Block Dessa page; The Verge; HN (410 pts).
Sources
Daily scan: 2026-10-24