OpenAI on Monday debuted a new chatbot with some impressive — or freaky, depending on your perspective — capabilities. With its ability to see, hear, and talk like a real person, OpenAI executives demonstrated how the latest ChatGPT can translate live conversations and tell bedtime stories in different voices.
OpenAI’s head of frontiers research, Mark Chen, asked the chatbot to tell “a bedtime story about robots and love.”
“Oh, a bedtime story about robots and love? I got you covered!” an enthusiastic feminine voice responded immediately. “Once upon a time, in a world not too different from ours, there was a robot named Byte. Byte was a curious robot. Always exploring.”
Chen interrupted, “I want a little bit more emotion in your voice, a little bit more drama.”
“Got it. Let’s add some drama,” said ChatGPT-4o, beginning its story again, but this time in a deeper voice with more intrigue.
Chen interceded once more, “No, no, no, ChatGPT, I really want maximal emotion, like maximal expressiveness, much more than you were doing before.”
“Understood. Let’s amplify the drama,” it responded, again repeating its opening line in a much louder, even-more-dramatic voice. OpenAI chief of technology Mira Murati asked the chatbot to switch to the story in a robot voice; it obeyed. Then fellow researcher Barret Zoph interjected, asking for a singsong voice. ChatGPT sighed and began to sing.
Murati Chen, and Zoph showed how ChatGPT-4o can translate a live conversation between Italian and French and how it can listen to you breathe and walk you through breathing exercises. The chatbot is twice as fast as GPT-4 turbo and available to free users.
“For the past couple of years, we’ve been very focused on improving the intelligence of these models, and they’ve gotten pretty good, said Murati. “But this is the first time That we are really making a huge step forward when it comes to the ease of use.”
Perhaps even more impressive than its real-time conversation skills is ChatGPT’s visual capabilities. The chatbot can look at someone’s face through a phone camera and tell them how it thinks they’re feeling from their facial expressions. It can look at a user’s computer screen through screen share and help solve a math problem.
Safety concerns
Murati said the chatbot’s capabilities, while a major improvement, pose “new challenges for us when it comes to safety.”
“Our team has been hard at work figuring out how to build in mitigations against misuse,” she said, adding that OpenAI is working with governments and companies across “all industries” to work toward its goals. She did not give specifics about safety concerns, OpenAI’s plan, or the nature of its partnerships.