OpenAI is introducing significant changes to ChatGPT, expanding its capabilities beyond text-based interactions. In the next two weeks, subscribers will gain access to features that enable voice prompts and image-based queries, with a wider release slated for the near future.
However, OpenAI acknowledges the inherent risks, including potential misuse for impersonation or fraud, associated with synthetic voices. To mitigate these concerns, access to the text-to-speech model will be tightly controlled and limited to specific use cases and partnerships.
The image search feature in ChatGPT resembles Google Lens, allowing users to snap photos and receive relevant responses. Users can refine their queries using the app’s drawing tool or by combining spoken or typed questions with images. This multimodal approach streamlines interactions, enabling users to iteratively refine their queries for more accurate results.
Despite its potential, image search introduces challenges, particularly when querying about individuals. OpenAI has intentionally restricted ChatGPT’s ability to analyse and make direct statements about people for reasons related to accuracy and privacy.
As OpenAI continues to enhance ChatGPT, it grapples with the balance between expanding its capabilities and addressing potential downsides. While the current releases cap the AI’s abilities intentionally, the ongoing evolution of voice control and image search will test the limits of these constraints. As ChatGPT becomes a more versatile and multi-modal virtual assistant, maintaining control over its use becomes increasingly complex.