ChatGPT has been writing text and software code since it debuted, but any refinement of your prompt required a complete rewrite. OpenAI has released a new feature called Canvas that provides a shared, editable page where ChatGPT can mimic a human worker and repeatedly edit or provide feedback on specific parts of the text and code you select.
A helpful way to think of Canvas is to imagine ChatGPT as your partner on a writing or coding project (you might even say “copilot” if you worked at Microsoft). Canvas works on a separate page outside the standard chatbot window, where you can ask the AI to write a blog post, code a mobile app function, and so on. Instead of reading through the result and asking for a change in tone or adjustment to the code, you can highlight the specific bits you want to change and comment on the type of edits you’re looking for.
So if you like what ChatGPT has written for your newsletter, apart from the introduction, you can highlight those paragraphs and say you want it to be more formal, or give a preview for the rest of the newsletter which is too short. You can do the same by editing your own text if you share some text and ask for longer text or if you use less complex language. The suggestions even extend to asking ChatGPT for emoji ideas.
The same general idea applies to getting ChatGPT to edit code on Canvas, whether AI-generated or human-written. You can ask ChatGPT to debug code, suggest improvements, or insert comments to make it more useful when sharing with real people. While emojis may not be relevant in coding software, you can use Canvas to ask ChatGPT to translate a program into another programming language, switching between the two depending on what’s most useful.
Canvas relies on the new GPT-4o model. For now, only ChatGPT Plus and Team subscribers have access, although this will soon be opened up to Enterprise and Education customers. OpenAI will also make it available to those who rely on the free version of ChatGPT, but not until the beta phase is complete.
Canvas opens in a separate window so you and ChatGPT can work on ideas side by side. In canvas, ChatGPT can suggest edits, adjust length, change reading levels, and provide inline feedback. You can also write and edit directly in canvas. pic.twitter.com/yHYVGtwJHVOctober 3, 2024
Blank canvas
Canvas is a logical step in OpenAI’s expansion of ChatGPT’s features. It looks a lot like a text version of the editing tools for AI-produced images created with OpenAI’s DALL-E models. Instead of highlighting part of an image and asking how to change it, Canvas focuses on the text. Often the text-based features come first and only later become multimodal in some form, so this is an interesting inverse of the standard release pattern.
The appeal to even the most casual ChatGPT users is obvious, as narrow-focus edits and suggestions are far more useful than the one-dimensional conversational approach of ChatGPT’s standard form. This is especially true when it comes to complex code or long text. Of course, this could reasonably raise the concerns of educators and others already concerned about misuse or over-reliance on AI-generated writing. It’s one thing for a student to ask ChatGPT to write an essay for him or her; it’s another when they can fine-tune the result to make it harder to spot signs of AI composition. Still, making ChatGPT a more active assistant and being able to task it with very specific problems will likely be a huge boon in more legitimate pursuits, especially if you need an editor who will never lose patience with you.
“People use ChatGPT every day for help with writing and coding. While the chat interface is easy to use and works well for many tasks, it is limited if you want to work on projects that require editing and revisions,” OpenAI explained in the announcement. “To make AI more useful and accessible, we need to rethink the way we interact with it. Canvas is a new approach and the first major update to ChatGPT’s visual interface since we launched two years ago.”