ChatGPT now accepts image and voice requests

ChatGPT, a chatbot from OpenAI, previously always required a text query to work. Now users of the paid version of the service will be able communicate with him in other ways: voice and pictures.

Voice queries are quite intuitive: you don't write a question, but click on the microphone icon and dictate it out loud. The system converts the request into text, generates a response and speaks it. The user can choose one of five voices for voicing answers. In general, this is similar to a conversation with “Alice” or Siri - the only difference is that the answers should be better due to an improved language model.

The image query function is similar to searching through Google Lens. The user sends a picture to the chatbot and ChatGPT will try to guess what the person meant and respond according to the request. Clarifying questions can be added by text or voice, which makes such a search more useful than the same “Google Lens”.

OpenAI notes that the developers have added a number of restrictions to image processing via ChatGPT. So, for example, the bot will not be able to evaluate and recognize people’s appearance in order to protect them from unwanted attention.

All users of the paid version of ChatGPT will have access to the new query options within two weeks, and access will be opened to non-subscription users shortly after that.