OpenAI’s Spring Event unveiled a new iteration of its popular AI product, ChatGPT. Called GPT 4o, it promises to work with speech, text and visual processing. OpenAI says the new model will be twice as fast as GPT 4, 50 percent cheaper and have 5 times the speed limits of the previous ChatGPT 4 Turbo. It will also offer a new desktop app for Mac, with a Windows version coming soon.
GPT-4o Voice Assistant is Finally Here and More AI Use Cases
In a very informal presentation (certainly compared to Apple’s more staged keynotes), Chief Technology Officer Mira Murati walked us through the announcement, then brought in a pair of research leaders, Mike Chen and Barrett Zoph, to lead live demonstrations.
The team demonstrated ChatGPT's new interaction style in real-time, pausing the chatbot and asking for various tone-of-voice adjustments. It seems like it now feels more intuitive to simply talk to ChatGPT in a normal conversation flow, with the voice output sounding much more natural when talking to the video hosts.
They also showed how GPT 4o can interact visually, with a phone camera pointed at some math problems and sentences. The chatbot was able to interpret and help the presenters with the math, and even interpret a heart as “love.” Most impressive was the intonation available in the GPT 4o “voice.” Although you can tell it’s computer generated, it seems to have a more natural rhythm that could make GPT 4o easier to interact with than previous iterations.
