The Day 4 OpenAI live stream introduced a new feature called Canvas, enabling users to collaborate with ChatGPT on writing and coding tasks, including real-time editing and feedback on documents. The demonstration showcased its applications in storytelling, academic writing, and programming, highlighting features like Python code execution and integration into custom GPTs for tailored user experiences.
The OpenAI live stream on Day 4 of their “12 Days of OpenAI Releases and Demos” focused on the introduction of a new feature called Canvas, which allows users to collaborate with ChatGPT on writing and coding tasks. The stream began with a brief introduction by Kevin, who mentioned the overwhelming demand for their recently launched product, Sora. He highlighted that Canvas would now be available to all users, integrating it into the main model, enabling Python code execution within Canvas, and allowing custom GPTs to utilize Canvas features.
The demonstration showcased how Canvas works, with engineers Lee and Alexi creating a Christmas story for children using the new feature. They illustrated the side-by-side view of the chat interface and the Canvas, allowing for real-time collaboration and editing. Users can provide feedback, request changes, and even add emojis to their documents, making the writing process more interactive and user-friendly. The ability to edit collaboratively in a separate document format was emphasized as a significant improvement over previous chat interactions.
In addition to storytelling, the stream highlighted Canvas’s capabilities for academic writing. One engineer demonstrated how to receive feedback on a physics essay, showcasing how ChatGPT could leave comments directly on specific sections of the text. This feature allows users to connect suggestions to their work more easily, making it a valuable tool for students and writers seeking constructive criticism. The ability to apply or reject suggested edits further enhances the collaborative writing experience.
The programming capabilities of Canvas were also a focal point of the stream. An engineer demonstrated how to debug Python code within Canvas, utilizing a built-in code editor with syntax highlighting and autocomplete features. The ability to run Python code directly in Canvas was a significant highlight, allowing users to receive immediate feedback on their code execution. This feature aims to streamline the coding process, making it easier for users to learn and troubleshoot their programming tasks.
Finally, the stream concluded with a discussion on integrating Canvas into custom GPTs, allowing users to create tailored experiences for specific tasks. The engineers explained how to set up custom instructions for GPTs to utilize Canvas automatically, enhancing the overall functionality of the tool. The excitement surrounding these new features was palpable, as the team expressed their eagerness to see how users would leverage Canvas for various creative and technical projects. The stream ended with a light-hearted Christmas joke, wrapping up a day filled with innovative announcements and demonstrations.