OpenAI introduces GPT-4o, featuring a voice mode for lifelike user interactions, with abilities to detect emotions, alter tone, and perform diverse tasks. The model’s potential integration with Siri signals a shift towards more expressive AI assistants.
OpenAI Unveils Sophisticated Voice Capabilities in GPT-4o AI Model
San Francisco, May 14, 2024 (Updated 2:08 PM ET)
OpenAI showcased the latest advancements in its AI technology with the introduction of GPT-4o, an updated version of ChatGPT. This new model features a voice mode designed to allow more conversational and lifelike interactions with users. The unveiling took place during an event in San Francisco and was marked by a series of live demonstrations, highlighting the AI’s ability to detect emotions, analyze facial expressions, and alter its tone and cadence.
The GPT-4o model can perform various tasks such as telling bedtime stories, responding in different voices, and even singing on command. This new feature has drawn comparisons to the virtual assistant Samantha from the 2013 film “Her,” directed by Spike Jonze. Sam Altman, OpenAI’s CEO, emphasized the model’s capabilities in a social media post, hinting at its resemblance to AI depicted in movies.
OpenAI’s voice assistant was demonstrated as capable of real-time translation between languages, and its voice adjusted to various tones, making it sound more humanlike. The AI’s ability to handle voice prompts natively, without converting them to text first, improves response times and conversational fluidity significantly.
Notably, Apple is reportedly in negotiations with OpenAI to integrate this technology into its existing AI assistant, Siri. The potential deal could be announced at Apple’s Worldwide Developers Conference in June, signaling a significant step forward in AI-enabled virtual assistance.
The GPT-4o model represents a shift towards more personable and expressive AI assistants, moving away from the detached, impersonal models like Siri and Alexa, and bringing a new era of interactive technology closer to reality.