New Delhi: Tavus, a San Francisco-based artificial intelligence research lab, has launched the first AI model to pass the video Turing test. The test is seen as a method to gauge whether a machine can show human-like intelligence in conversation, and Tavus claims its latest AI model Griffin has passed it.
In a post on X Thursday, the company said that 48 per cent of people who talked to the interaction-based model thought it was human.
“48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA’s benchmark for full-duplex AI video,” Tavus posted.
It described the AI model as the first-ever ‘Human Interaction Model (HIM)’, an LLM that holds face-to-face conversations in real time. In a thread of posts featuring Tavus founder Hassaan Raza and others, the company showed various ways the model can be used.
The Turing Test was first proposed by British mathematician Alan Turing in his 1950 paper ‘Computing Machinery and Intelligence’. He originally called it the Imitation Game, and the test is considered to be outdated by many today. While modern AI can trick people in a chat, it does not truly understand what it is saying.
Do we need machines to be more ‘human’?
Created in 2020, Tavus focuses on “human computing”, teaching machines to ‘see, hear, and respond’ with emotional intelligence. It builds interactive digital human avatars, Griffin being its newest model.
The company’s announcement was met on X with widespread criticism, with many saying that it would be misused to scam elderly people. Among its staunchest critics is US Senator Bernie Sanders, who called the development “insane”.
“In a new experiment, nearly half the people on a video call thought they were speaking to a human. They weren’t. It was a new AI model. That is INSANE. Does anyone in America think it is a good idea to race forward to build machines that replace us? PAUSE AI NOW,” Sanders wrote.
In what seems to be a response to public suspicion around the Griffin, Raza posted a write-up on X Saturday as a way to ‘take a moment and share the company’s thoughts’.
“Griffin, our latest model, isn’t publicly available yet. Yesterday’s announcement was a limited research preview to demonstrate what the model is capable of. With AI progressing so quickly, we prefer to share breakthroughs openly and in real-time as we work on a safe public release,” Raza said.
He listed several potential uses for the new technology, claiming that the model could be used as a tutor, health expert, or language coach.
“We’re working with partners on safeguards and systems for disclosure, as well as inviting discussions with officials around wider regulation and safe use. People will always know they’re interacting with AI, while providing an interface that removes the need to ‘speak computer’,” Raza wrote.
Also read: From Raja Bhoja to Mughals, how archaeologists traced Bhojshala’s changing past