ChatGPT Announces AI Chatbot Can Now Speak

Things just got a little more talkative in the AI universe. The company behind ChatGPT announced that its chatbot will now have the ability to speak. Users will now be able to have a back and forth conversation with ChatGPT, and that’s not all. The artificially intelligent bot will be able to converse in five different languages. It’s all part of OpenAI’s new upgrade to its iOS and Android apps. Let’s bring in Joanna Stern for more on this. She’s the NBC News personal technology analyst and the senior personal technology columnist for the Wall Street Journal. Joanna, always good to have you with us. So walk us through this new update. What do we know about these new vocal settings and how does OpenAI determine even how these voices are made?

Yeah, so we’ve had on one end systems like ChatGPT and Google’s Bard, which are large language models that are really generative AI. They can talk and speak a lot like humans. The type of text, the type of conversation that gets spit out in text, that’s what it seems like. On the other end, we’ve had systems like Siri and Alexa that are just voice assistants, that just have a voice and you can prompt them to do things and this is the start of seeing those things come together. Now we are able to talk to ChatGPT. It has a voice. I have spoken to it. It sounds very human-like. And you are still getting that text response or it’s doing text to speech. It is reading the text response that ChatGPT gives you.

So I do want to hear what this sounds like. I know we do have that clip of you having a conversation with ChatGPT. First, let’s take a listen.

Pretend I’m your best friend, ask me how my day was.

Of course friend, how was your day? Did anything interesting or special happen? I’m here to listen.

I shot a video at work.

That’s cool.

So not a rapid fire conversation, but what did you learn about the app and what is ChatGPT possibly learning about us through this, albeit slowly?

Yes, absolutely. So as I said in my piece, there are a lot of signs this thing is more like humans. And in terms of the cadence and the intonation, and you heard that in that voice. It did not sound robotic like you might think of an Alexa or Siri, but on the other hand it is very slow and it is clearly still a machine and there are some kinks they have to work out there, though OpenAI said that they have already solved some of those and that I was testing a beta app. But to your point here, people are going to be able to converse with this if they pay for ChatGPT Plus, which is the subscription version, that’s about $20 a month. And yeah, you’ll be able to now talk to ChatGPT very much like an Alexa or a Siri.

To be fair, there are days when I respond that slowly to people too. So this upgrade also includes a new camera feature, I understand. How does that enhance the app’s abilities?

So this is very cool. Currently, you’ve been able to talk to ChatGPT, use your voice or type in words. Now you can actually take a photo or upload a photo and use that to prompt ChatGPT. So one example as you’re seeing right here is I took a picture of a leaky faucet in my garage, as you can see, or sorry, hose. As you can see, I didn’t really know how to solve this, so I asked “How can I fix this?” I didn’t say what it was, and I just put this into the system and automatically it came back with some advice. So albeit not great advice on how to fix the hose. And so what’s really cool there is that you’re not having to describe something.

All right, very interesting. All right. It keeps evolving ever so quickly. Joanna Stern, we appreciate you keeping us up to date on all of this.