Large language models and other generative AI have the potential to greatly improve scientific productivity. But to fully realize this potential, the way that scientists interact with these tools will need to move beyond simple chat interfaces and copy-paste workflows.
We're working on the next version of Stencila. It brings generative AI to scientific authoring in ways that improve usability, reliability, and transparency. This includes providing LLMs with contextual information (including from the document, data, and the literature) to improve accuracy and relevance, a library of task-specific prompts, and routing tasks to the LLM that performs best.
Stencila's VS Code extension in development (editing this document). Coming to the VS Code Marketplace soon!
