OpenAI, the creator of ChatGPT, unveiled on Monday its latest AI innovation named GPT-4o, boasting lifelike voice interaction and the ability to seamlessly engage with text and images, marking yet another stride in its quest to lead the AI field.
The newly added audio functionalities empower users to engage in real-time conversations with ChatGPT without any discernible delay, even allowing interruptions mid-conversation—an achievement long deemed challenging for AI voice assistants, as demonstrated by OpenAI researchers during a live presentation.
"It's reminiscent of AI depicted in movies... Communicating with a computer has always felt somewhat artificial to me, but now it's remarkably natural," remarked OpenAI CEO Sam Altman in a blog entry.
Backed by Microsoft, OpenAI is confronted with mounting competition and the imperative to expand the user base of ChatGPT, its widely acclaimed chatbot celebrated for its human-like text generation and exceptional code writing capabilities.
During the live event, OpenAI researchers showcased ChatGPT's latest voice assistant features. In one demonstration, ChatGPT leveraged its visual and auditory prowess to guide a researcher through solving a mathematical problem on paper.
In another showcase, the GPT-4o model exhibited its prowess in real-time language translation.
The demonstrations presented by OpenAI bordered on science fiction, with ChatGPT and its conversational partner engaging in flirtatious banter at one juncture. Upon being complimented by the researcher on its utility and brilliance, ChatGPT quipped, "Oh stop it! You're making me blush!"
Following the presentation, Altman posted a cryptic message referencing the 2013 film "Her" by Spike Jonze, where a man develops romantic feelings for his AI assistant voiced by Scarlett Johansson.
OpenAI's Chief Technology Officer, Mira Murati, disclosed during the event that the new model would be available for free due to its cost-effectiveness compared to previous iterations. Premium users of GPT-4o will enjoy enhanced capacity limits compared to free users, Murati added. The integration of the GPT-4o model into ChatGPT is slated to roll out over the next few weeks, the company confirmed.
Moreover, free users of ChatGPT now have access to a "browse" feature enabling ChatGPT to provide up-to-date information from the web, Murati informed Reuters post-event. The company clarified that it has no intention to monetize free users through advertisements.
ChatGPT made waves shortly after its late 2022 launch, becoming the fastest application to amass 100 million monthly active users. However, global traffic to the ChatGPT website has experienced fluctuations over the past year, only recently rebounding to its May 2023 peak, according to analytics from Similarweb.
OpenAI timed its announcements a day ahead of Alphabet's annual Google developers' conference, where the tech giant is anticipated to unveil its own AI-related innovations. Reuters previously reported that OpenAI intended to introduce an AI-driven search product but opted to delay the announcement, according to a reliable source familiar with the matter.
Alphabet shares dipped 0.4% on Monday afternoon, following an earlier decline of nearly 3%. Similarly, Microsoft shares experienced a 0.2% drop.
