OpenAI is taking conversational AI to the next level with ChatGPT now having voice and image capabilities. This advancement will enhance how users interact with AI, making conversations more intuitive and informative.
Voice Conversations with ChatGPT
Starting soon, ChatGPT will enable users to engage in voice conversations with their AI assistant. Whether you’re on the go, settling a family debate, or just seeking a friendly chat, ChatGPT’s new voice feature is at your service. To initiate voice conversations, users can navigate to Settings → New Features on the mobile app and opt into this exciting feature. With five distinct voices to choose from, users can tailor their AI interactions to their preferences.
The new voice capability is powered by a state-of-the-art text-to-speech model, which can generate remarkably human-like audio from mere text inputs. Each voice is meticulously crafted through collaboration with professional voice actors, ensuring a high-quality conversational experience. Additionally, Whisper, an open-source speech recognition system, transcribes spoken words into text, facilitating seamless interactions.
Conversations About Images
ChatGPT now has the ability to process images, making it a versatile tool for various tasks. Users can troubleshoot issues, plan meals by inspecting fridge contents, or analyze complex data graphs using this newfound feature. The mobile app even includes a drawing tool to highlight specific details within images.
The image understanding capability is powered by multimodal models GPT-3.5 and GPT-4, which apply their language reasoning skills to interpret a wide array of images, including photographs, screenshots, and documents with both text and images.
Gradual Deployment for Safety and Improvement
OpenAI’s focus on building safe and beneficial artificial general intelligence (AGI) is reflected in its gradual deployment strategy. With the introduction of features like voice and image capabilities incrementally, OpenAI can refine risk mitigations and prepare users for more powerful systems in the future.
Voice and Image: Potential and Responsibility
While the new voice and image capabilities offer immense potential for creativity and accessibility, they also raise new challenges. OpenAI is taking precautions to prevent misuse. Voice chat, for instance, involves collaboration with voice actors and partner companies like Spotify to ensure responsible usage.
Balancing Utility and Privacy in Vision
The vision-based capabilities aim to assist users in their daily lives by interpreting images, but OpenAI is committed to safeguarding privacy. Technical measures have been implemented to limit ChatGPT’s ability to analyze and make statements about individuals. Real-world usage and feedback will help improve these safeguards.
Transparency and ChatGPT Limitations
OpenAI is transparent about ChatGPT’s limitations, especially for non-English languages and high-risk use cases. Users are encouraged to verify information from specialized topics and use caution when relying on ChatGPT for certain purposes.
Voice and image capabilities will first roll out to Plus and Enterprise users, with plans to expand access to developers and other user groups in the near future.