What is Dessa, Block's experimental voice AI?
Jack Dorsey’s Block has released an experimental AI model called Dessa, and the company is showing off its rough edges as much as its polish. Block says Dessa was trained “from scratch on nothing but human conversation.” To demonstrate it, the company shared chats between the model and real people, some of which are smooth and some of which are noticeably strange.
In one exchange shared by Dessa lead Ragavan Thurairatnam, the back-and-forth flows mostly naturally. The model does stumble at the start, though. For some reason, it initially mistakes the speaker for someone named “Ronaldo.”
Block also posted a video of Dessa playing a guessing game over voice. The Dessa website includes other examples. In one, a conversation about Christopher Nolan’s films takes an unexpected turn when Dessa goes off on a tangent about Bryan Singer’s Superman movies. In another, a person tells Dessa about taking it on a surprise trip to Paris, and the model responds with what the website describes as “teary gratitude.”
A third conversation pits Dessa against GPT-Live-1, a voice model from OpenAI. The website says Dessa comes across as more natural in that exchange.
How was Dessa trained?
The Dessa team describes the model as a “true speech-to-speech model, and everything it knows came from human speech.” According to the website, it can run locally on a MacBook Pro M4 Max “or better.”
Why is Block building Dessa?
The team lists three reasons for building it: “curiosity (aka fun), better understanding between humans and AI, and safety.” The starting point is a comparison to how children learn. “Babies learn to talk without ever seeing a word written down,” the team wrote. “We wanted to see if a model could too. It can. It’s just very hard.”
The team also argues that speech is the most natural way to interact with an intelligence, and that today’s voice AI wastes that advantage. In its view, voice systems flatten what people say into text and cannot keep up with human timing.
The safety argument is the most pointed. The team says AI is gaining power and independence while remaining “nearly deaf to the emotion in our voices.” It offers a scenario to illustrate the problem: “Imagine being in serious pain while a voice AI has no idea and just keeps going. That’s the safety problem. Dessa is a step toward fixing it.”
Block says it will give “a small group of AI safety researchers” early access to the model. The company has not said when, or whether, Dessa will be released more widely.
Dieser Artikel wurde mit Unterstützung von KI verfasst.
News Factory APP - agentische News für besseres SEO & AEO.