ChatGPT now has voice features for all users, Anthropic has updated its Claude chatbot and Stability AI is moving into the generative AI video market.
The AI race continues as multiple companies in this sector have announced new features for their services.
OpenAI has announced that its flagship product – ChatGPT – now has voice features available to all free users. This was previously a feature that was limited to the chatbot’s premium subscribers.
The company made the announcement on X, with a video clip that contains a joke about the recent drama occurring within the company. A staff member tells ChatGPT that it has “been a long day” and asks how many pizzas she should order for 778 people.
OpenAI has had a chaotic few days since the company’s board suddenly fired Sam Altman from his position as CEO. There were back-and-forth developments over the weekend, including Altman being offered a job leading a Microsoft AI research team.
But it appears OpenAI has gone back on its decision and has reached an agreement “in principle” to have Altman take the reins of the company again, with a new board initially comprising of Larry Summers, Adam D’Angelo and Bret Taylor as chair.
New features for Anthropic’s Claude
Amid the chaos at OpenAI, one of its rivals – Anthropic – has released a new version of its ChatGPT-challenger Claude. The company said these changes brings new capabilities and significant reductions in hallucinations – a term to describe when an AI chatbot makes mistakes.
Anthropic said the latest version of its Claude chatbot lets users relay double the amount of information they could with the earlier model. This lets users upload large literary works or detailed financial statements, according to the company.
“By being able to talk to large bodies of content or data, Claude can summarise, perform Q&A, forecast trends, compare and contrast multiple documents and much more,” Anthropic said in a blogpost.
Anthropic also claims that the significant decline in hallucinations will let enterprises deploy AI across their operations with “greater trust and reliability”. The company claims its chatbot is more likely to say it doesn’t know an answer to a certain question, rather than provide false information as a response.
The updates include a new pricing model in order to “improve cost efficiency” for its customers across the various Claude models. Claude Instant comes at the cheapest price overall, while Claude 2.0 and Claude 2.1 have the same pricing models.
Stability AI goes to video
Meanwhile, Stability AI – the company behind the text-to-image generator Stable Diffusion – has released its first foundational model for generative AI video content. This model is based on Stable Diffusion and is available in research preview.
The company said this model is state-of-the-art and represents a “significant step” on its journey toward creating models “for everyone of every type”.
“Stable Video Diffusion is released in the form of two image-to-video models, capable of generating 14 and 25 frames at customisable frame rates between three and 30 frames per second,” Stability AI said in a blogpost. “At the time of release in their foundational form, through external evaluation, we have found these models surpass the leading closed models in user preference studies.”
The company has released the code for this new AI model on its GitHub repository and has also released a research paper detailing the capabilities of this system. Users can also sign up to a waiting list for an “upcoming web experience featuring a text-to-video interface”.
Stability AI is best known for its text-to-image generator, which is in a competitive market against tools such as OpenAI’s Dall-E models. But the company has been shifting into new sectors recently, including audio.
Last week, Ed Newton-Rex – who led the audio team at Stability AI – resigned from his position and claimed he disagrees with the company’s opinion that training generative AI models on copyrighted works is “fair use”.
10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.