inputs/markdown/openai_canvas_intro.md

Introducing Canvas

What is Canvas?

• A new interface for working with ChatGPT on writing and coding projects
• Opens in a separate window, allowing collaboration on a project

Better Collaboration with ChatGPT

• Limited by chat interface when working on editing and revisions
• Canvas offers a new interface for this kind of work

How Does Canvas Work?

• You control the project, directly editing text or code
• Menu of shortcuts to ask ChatGPT to adjust writing length, debug code, etc.
• Can restore previous versions of your work using the back button in canvas

Writing Shortcuts in Canvas

• Suggest edits: Inline suggestions and feedback
• Adjust the length: Edits document length to be shorter or longer
• Change reading level: Adjusts reading level from Kindergarten to Graduate School
• Add final polish: Checks for grammar, clarity, and consistency
• Add emojis: Adds relevant emojis for emphasis and color

Coding in Canvas

• Easier to track and understand ChatGPT's changes
• Planned improvements to transparency into these kinds of edits
• Coding shortcuts:
  • Review code: Inline suggestions to improve your code
  • Add logs: Inserts print statements to help you debug and understand your code
  • Add comments: Adds comments to the code to make it easier to understand
  • Fix bugs: Detects and rewrites problematic code to resolve errors
  • Port to a language: Translates your code into JavaScript, TypeScript, Python, Java, C++, or PHP

Training the Model for Collaboration

• Trained GPT-4o to collaborate as a creative partner
• Understands broader context to provide precise feedback and suggestions

Core Behaviors of the Model

• Triggering the canvas for writing and coding
• Generating diverse content types
• Making targeted edits
• Rewriting documents
• Providing inline critique

Improving Canvas Decision Boundary

• Improved correctly triggering the canvas decision boundary to 83% and 94% respectively for writing and coding tasks

Canvas Edits Boundary - Writing & Coding

• GPT-4o with canvas performs better than a baseline prompted GPT-4o by 18%

Training the Model to Generate High-Quality Comments

• Used human evaluations to assess comment quality and accuracy
• Outperforms zero-shot GPT-4o with prompted instructions by 30% in accuracy and 16% in quality

What's Next?

• Rethinking how we interact with AI requires updates like Canvas
• Canvas is in early beta, and rapid improvements are planned