OpenAI expands ChatGPT with real-time video feature

OpenAI has officially released the long-awaited real-time video feature for ChatGPT, seven months after it was first demonstrated. During a livestream presentation, the company showcased the new "Advanced Voice Mode" feature with visual capabilities. Users of ChatGPT Plus, Team, or Pro can now point their smartphones at objects and receive responses from ChatGPT almost in real time. With the new feature, ChatGPT can not only recognize real objects but also analyze and describe content on a device's screen. This may include explanations of various menu settings or solutions to mathematical problems. Access to Advanced Voice Mode with visual capabilities is achieved by tapping the voice icon next to the chat bar and then the video icon at the bottom left. To share content, the screen can be shared via selection through the three-dot menu. However, not every user will enjoy these new capabilities. While the rollout of the feature has already begun, ChatGPT Enterprise and Edu subscribers will have to wait until January. Users from the EU, Switzerland, Iceland, Norway, and Liechtenstein must also wait indefinitely. A high-profile test of the new feature was recently shown on the "60 Minutes" show on CNN, when OpenAI President Greg Brockman challenged Anderson Cooper with an anatomical drawing task. ChatGPT recognized the sketches immediately, but also revealed weaknesses in mathematical tasks, indicating possible inaccuracies. Originally, the feature was supposed to be available in the spring but was postponed multiple times. When the voice-only version was released for some users in the fall, the visual component was still missing. As OpenAI intensifies its efforts, competitors like Google and Meta are also developing similar technologies. Google has just made its "Astra" project, a real-time video analysis AI, available to selected testers on Android.