— OpenAI’s latest stride in artificial intelligence marks a significant advancement, showcasing capabilities reminiscent of science fiction. The update to its AI model can now mimic human cadences in verbal responses and is equipped to discern people’s moods, echoing themes seen in the 2013 Spike Jonze film “Her.”

Though the allure of the new model may not reach the romantic heights depicted in the movie, OpenAI asserts that it boasts improved speed and cross-modal reasoning abilities, capable of analyzing text, audio, and video in real time.

Dubbed GPT-4o, or “omni,” the AI will power OpenAI’s ChatGPT chatbot, with availability extending to users, including those on the free tier, in the forthcoming weeks, as announced during a succinct live-streamed update. CEO Sam Altman, absent from the event, cryptically shared “her” on social media platform X.

In a demonstration featuring Chief Technology Officer Mira Murati and other executives, the AI engaged in real-time conversation, infusing emotions — notably, “more drama” — into its voice upon request. Furthermore, it provided step-by-step guidance for solving mathematical equations and tackled complex software coding dilemmas on-screen.

The AI also ventured into assessing emotional states, interpreting a person’s mood from a selfie video (concluding happiness from a smile), and facilitated multilingual conversations by translating between English and Italian.

Gartner analyst Chirag Dekate observed that the update, albeit brief, may suggest OpenAI is striving to keep pace with industry frontrunners.

“Many of the showcased demos and capabilities by OpenAI felt familiar, as advanced versions had been exhibited by Google in their Gemini 1.5 pro launch,” Dekate remarked. “Despite OpenAI’s initial advantage last year with ChatGPT and GPT3, comparative assessments reveal emerging capability gaps when stacked against peers, particularly Google.”

Google’s forthcoming I/O developer conference, scheduled for Tuesday and Wednesday, is anticipated to unveil updates to its own Gemini AI model.

Copyright 2024 The Associated Press. All rights reserved.


Leave a Reply

Your email address will not be published. Required fields are marked *

Seriously a household fave! We can’t get enough of this one!. FULL RECIPE 💬👇. OpenAI launches GPTo, improving ChatGPT’s text, visual and audio capabilities GameStop shares jump for a second day, but are well off the highs as meme enthusiasm starts to fade Cardiac Celtics Part III: first ‘clutch time’ playoff game wasn’t fun