TechNews May 15th - OpenAI released the GPT-4o model in the early hours of yesterday, which can be seen as an upgraded version of GPT-4, capable of real-time reasoning with audio, visual, and text content, and can interact more seamlessly with ChatGPT.
OpenAI President and Co-Founder Greg Brockman then released a hands-on demonstration, allowing two devices running GPT-4o to engage in voice or video interactions.
Before the demonstration began, Brockman expressed his hope: to let two AI chatbots converse with each other. He told Chatbot A about his expectations and allowed Chatbot B to access the camera to observe the surrounding world.
Chatbot B accessed the front camera, providing a clear description of the user's attire and surroundings. Chatbot A could then converse with it and ask questions, including moving the camera and the content it saw.
In the conversation, the new version of ChatGPT not only understands what is happening in the camera through visual AI capabilities, but also conveys that understanding through voice interaction for a more enriching and engaging experience. It also supports interruptions and insertion of dialogue, along with context memory capabilities. TapTechNews has attached a demonstration video below: