news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Produced by Huxiu Technology Group

Author: Yu Yang

Editor: Miao Zhengqing

Header Image｜Visual China

News on August 1,OpenAI NewChatGPTAdvanced Voice Mode, however, is limited to a select few who subscribe to ChatGPT Plus.

In fact, OpenAIGPT-4oThe feature was shown at the launch event, a responsive AI assistant character, but was later delayed for safety reasons.

In a livestreamed demonstration in May, OpenAI engineers and CTO Mira Murati gathered around the phone to show off new features. They first encouraged the assistant to be more expressive while making up a bedtime story, then suddenly asked it to switch to a robotic voice, and finally asked it to end the story with a singing voice. Not only that, they also asked the assistant to see what the phone's camera saw and had it react to what was visible on the screen. The assistant was also able to be interrupted while speaking and respond without being prompted to continue while acting as a translator.

At the time, OpenAI CEO Sam Altman mysteriously posted just one word on X: "Her."Altman once said frankly that "Her" was his favorite movie.

It was September 2023, and against a leafy backdrop on the Yerba Buena Theater stage, Altman met with Salesforce CEO Marc Benioff and, with a packed audience, talked about everything from generative AI to his favorite sci-fi movie, Her.

“I love Her. The things that Her does right — like the whole interaction model of how people use AI — are incredibly prophetic,” Altman said.

"Her" is a science fiction romance film directed by Spike Jonze, shot in 2013. The film tells the story of humans andDigital HumanThe protagonist, Theodore, is a letter writer who is delicate and profound and can write the most touching letters. He has just ended his marriage to his wife Catherine and is still immersed in heartbreak. By chance, he comes into contact with the latest artificial intelligence system OS1, whichDigital AvatarSamantha has a charming voice, gentle and humorous. Theodore and Samantha soon discovered that they were very compatible, with two-way needs and desires, and the human-machine friendship eventually developed into a love beyond worldly understanding.

It is not difficult to find that the film focuses on topics such as human loneliness and the future relationship between humans and artificial intelligence. It combines novel science fiction settings with traditional romantic atmosphere. The charm of technology explodes frame by frame in the interaction between real people and virtual reality, and accurately hits the G-spot of a group of AI engineers.To a certain extent, it can be said that "Her" is the starting point of the dreams of many dreamers who are now driving the continuous interpretation of artificial intelligence.

It is worth mentioning that the wonderful dubbing performance of Samantha in "Her" comes from Scarlett Johansson.ScarlettShe won the Best Actress Award in Rome just for her voice without even showing her face.Ultraman was also fascinated by it.

As early as May, Altman invited Scarlett to dub the ChatGPT 4.0 system. After being rejected, Altman persevered and asked Scarlett to reconsider.

The subsequently released voice - Sky - also raised the risk of litigation due to its high similarity to Scarlett's voice, and became one of the "safety reasons" for Open AI's delay in releasing the product.

However, during the month of delayed release, OpenAI also made a step forward.

During OpenAI’s demo event, OpenAI employees were still able to interrupt the chatbot and ask it to tell a story differently, and the chatbot adjusted and responded with ease.But in ChatGPT's new mode, there are only four preset voices that it created with voice actors, and ChatGPT cannot imitate other people's voices, including individuals and public figures.

OpenAI spokeswoman Taya Christianson said the company worked with more than 100 external red teams (people who try to hack technology to find weaknesses) to test the voice model's capabilities. It also added new filters that can identify and block certain requests to generate music or other copyrighted audio, showing OpenAI's efforts on safety issues.

What’s interesting is that there is a conversation like this at the end of the film.

Theodore asked Samantha, "How many people are you talking to at the same time? How many people are you in love with at the same time?"

Samantha responded: "8361,641."

While Theodore was intoxicated by Samantha's tenderness, sensuality and thoughtfulness, Samantha, as the OS, had to deal with the relationship with other users at the same time. This love between man and machine eventually reached BE with this as the node.

Compared with Zack's dream of trying to intervene in the real world, Ultraman's persistence in the dream of "Her" may reveal the difference between OpenAI and Meta in some dimension.