OpenAI created a sensation by unveiling its new multimodal language model GPT-4o, which can reason about audio, vision and text simultaneously. As for the new associated voice mode, it gives the impression of truly chatting with someone close to you, capable of empathy, rather than with a simple assistant.
This new GPT-4o (“o” for “omni”) model leads to much more natural human-machine interaction than with GPT-4. In addition to being more responsive, the voice assistant is now able to adapt to your mood and even sing if necessary. The capabilities of the ChatGPT mobile application are increased tenfold, with this model capable of precisely describing and interacting with any image on the screen.
ChatGPT's voice mode opens up to new use cases which can prove very practical. As a reminder, you can communicate with the voice assistant available in the ChatGPT mobile application by pressing the headset-shaped logo at the bottom right of your smartphone screen.
A personal coach
Before a professional interview or a romantic date, you can ask ChatGPT if you are ready, by asking about your general appearance or your outfit. All you have to do is film yourself with your smartphone camera and ask: “Am I presentable? Do I look professional?” In addition to commenting on your appearance, it will also be able to recognize your emotions, especially if you are stressed. In this case, he can offer personalized advice or suggest simple exercises to relax.
An automatic translator
It is possible to ask to translate what you say into any language. It's practical for revising a language wherever you are or simply for making yourself understood if you are abroad. You can therefore issue an instruction: “Translate everything I am going to say from now on into English”. At any time, you can also request to switch to another language if you wish. Even stronger, you can communicate with someone who does not speak your language by asking for example: “Translate into English every time you hear me speak in French and translate into French every time you hear him speak in English”. The result then turns out to be astonishing.
A home teacher
Generally speaking, the voice assistant can replace a teacher when any exercise is presented to it. You still need to give him the right instructions. For example, a student can point their camera at a math problem and ask ChatGPT to help them solve it in an intelligent way: “Help me solve this problem, but don't tell me the answer. Ask me questions instead that will help me better understand how to get there on my own.” This method can also be adapted to multitasking on an iPad for example, with ChatGPT open and the exercise in question displayed next to it.
A referee
ChatGPT can serve as a referee for very fun little visual games, like rock-paper-scissors. If there are two of you and you stand in front of the camera, you can then play and the voice assistant can systematically choose the winner. Just give him instructions: “We are now going to play rock-paper-scissors, will you tell us who wins after each action?”
A friend of animals
Using your smartphone's camera, you can introduce your dog or cat to ChatGPT's voice assistant, and ask for practical advice: “This is Médor, do you recognize his breed? What is Could I suggest an activity for him today?
OpenAI promises that GPT-4o will soon be available for free to all ChatGPT users. For the moment, only owners of a paid account can benefit from it. As for the new ultra-responsive version of voice mode, it will be launched a little later this year.
