OpenAI announced Monday that it was launching its new flagship AI model, called GPT-4o, as well as updates that included a new desktop service and advancements in its voice assistant capabilities.
Chief technology officer Mira Murati appeared on stage to a cheering crowd at the OpenAI offices, touting the new model as a step forward in AI. The new model will bring the faster and more accurate GPT-4 AI model to free users, where it was previously reserved for paid customers.
“We are looking at the future of the interaction between us and machines,” Murati said. “We think GPT-4o is really changing that paradigm.”
Among the updates OpenAI revealed on Monday were improving the quality and speed of ChatGPT’s international language capabilities, as well as the ability to upload images, audio, and text documents for the model to analyze. The company said it would roll out features gradually to ensure they are used safely.
The event also included a live demonstration of the model’s new voice capabilities, with two OpenAI research leaders speaking to an AI voice model. The voice assistant generated a bedtime story about love and robots, and the researchers told him to speak with a variety of different emotional and vocal inflections. Another demonstration used a phone’s camera feature to show the AI model a mathematical equation and then ChatGPT’s voice mode told them how to solve it.
At one point in the demonstration, a researcher asked the AI model to read his facial expression and judge his emotions. ChatGPT’s voice assistant assessed that he looked “happy and cheerful with a big smile and maybe even a touch of excitement.”
“No matter what, it seems like you’re in a very good mood,” ChatGPT said in a cheerful, feminine voice. “Would you mind sharing the source of those good vibes?”
OpenAI was rumored last week to launch a search product that would compete with Google, but Reuters reported that the company had delayed revealing that project. CEO Sam Altman denied that Monday’s announcement was a search engine following reports, but he tweeted on Friday that “we’ve been working hard on some new things that we think people will love! It seems magical to me.”
Aside from minor glitches or inadvertent responses, most of the demo ran smoothly and the features worked as intended. Although Murati stated that OpenAI was taking steps to prevent its new voice capabilities from being misused, the event provided few details about how the company would address safeguards around facial recognition and audio generation.
after newsletter promotion
OpenAI is also reportedly close to closing a deal to incorporate its generative AI features into Apple’s iPhone operating system. according to Bloomberg. Apple has been trying to shift resources toward AI as it lacks a publicly available generative AI product similar to competitors like Google’s Gemini flagship. Its voice mode seemed to potentially offer more advanced features than Apple’s Siri voice assistant can offer.
As OpenAI expands its products and seeks partnerships, it is also facing a number of lawsuits from publishers and media outlets over alleged copyright violations. News organizations like the New York Times accuse OpenAI of illegally training its AI models in its work without compensation or consent, seeking damages that could run into the billions.