1. Home >
  2. Computing

ChatGPT to Begin Allowing Photo and Voice-Based Queries

These new features will begin rolling out to paid customers first, with users of the free version getting them sometime in the near future.
By Josh Norem
ChatGPT interface
Credit: Emiliano Vittoriosi/Unsplash

ChatGPT's arrival on the scene in late 2022 likely made the C-suite at Google a bit nervous. Here was this upstart chatbot providing answers to questions for people, without ads, in a way that was—dare we say—delightful? It was a clear threat to Google's search business. However, it has remained a mere curiosity until now for most people due to its limited access to the internet and Google's stranglehold over the industry. Now OpenAI, which runs ChatGPT, is firing another shot across Google's bow by integrating voice and photo queries into its popular chatbot.

OpenAI has announced it's rolling out a new version of ChatGPT shortly that will let you just ask the bot a question with your voice. The service will convert it to text, feed it to ChatGPT, convert the answer to audio, and read it back to you, according to The Verge. You can choose from five voice options, which the company says provide "human-like audio." It sounds like it'll offer more than five voices in the future, but this is just a starting point for the company on its speech-to-text journey.

You can also just upload a photo, which seems dangerous, but perhaps it already has some kind of nudity detection built-in. It also begs the question of what it would say about a photo of a person. OpenAI says it limited ChatGPT's ability to comment about people, but clearly, there are grey areas. For example, you could upload a photo of a past president and ask about their legacy or a blurry image of your neighbor, which is painful even to type.

OpenAI says it built this feature so you can upload an image and then ask the bot questions about it to help it refine the answers it gives. In the example above, the user posts a photo of their bike and asks ChatGPT how to lower the seat. The bot replies with instructions, and they follow up with another picture of the seat's ride-height mechanism for more guidance. Google Lens offers a similar service currently but without the back-and-forth refinement of queries.

The company says it will begin rolling out these features for paying customers in the next two weeks, with free user access later. These features will open up new concerns about privacy, accuracy, and whether the bot is "hallucinating" like these services are known to do occasionally.

Tagged In

Artificial Intelligence

More from Computing

Subscribe Today to get the latest ExtremeTech news delivered right to your inbox.
This newsletter may contain advertising, deals, or affiliate links. Subscribing to a newsletter indicates your consent to our Terms of use(Opens in a new window) and Privacy Policy. You may unsubscribe from the newsletter at any time.
Thanks for Signing Up