Overview
AI Chat is a flexible, advisory tool for qualitative research. AI Chat acts like a research assistant to help you:
Generate ideas for new codes or subcodes
Summarize a collection of snippets for a code
Write or improve code descriptions
Find quotes or illustrative examples from your data
Identify snippet outliers
Explore and analyze your coded data
Always combine AI suggestions with your own knowledge of your study, data, and research goals to ensure accurate and meaningful analysis. Use AI Chat as part of an iterative, human-reviewed workflow, not replacement.
Note: To access AI Chat, you must join Delve’s AI Beta program by accepting the Beta Terms.
Where to Access AI Chat
Click AI Chat in the left-hand menu in Delve. The AI Chat panel will open.
From here, you can ask questions about your coded snippets, transcripts, or codes.
Note: AI Chat works best after an initial round of broad coding. Consider using Transcript AI first for high-level analysis , then refine and explore your data with AI Chat.
Using AI Chat
Step 1: Determine Scope
You can select which snippets and what transcripts will be used by AI Chat when answering.
Codes filter: AI Chat will look only at snippets coded under the selected code(s).
Transcripts filter: AI Chat will include all coded snippets in the selected transcript(s).
Note: By default, AI Chat considers all available coded snippets, but narrowing your filter improves answer relevance. For large datasets, it’s recommended to utilize filters for better results as AI models have limited memory.
Step 2: Ask Your Question
Type your question into the chat box. Examples:
"Give me ideas for themes"
"Summarize snippets for [code name]"
"Suggest subcodes for [code name]"
Step 3: Refine and Follow Up
Iterate with AI Chat to refine answers:
Ask for clarifications: "Explain why you grouped these snippets this way"
Change the format: "Give me the answer in a table with snippet citations"
Request instructional summaries: "Provide an instruction on what kind of snippet belongs in this code"
Connect to theory: "Situate this code in [theoretical framework]"
Step 4: Optional Coding
AI Chat is meant to be advisory and help you brainstorm, but you do have the option to apply suggested codes.
To code snippets suggested by AI Chat, click See x supporting snippets.
Select the snippet and click the appropriate code.
Best Practices
Be familiar with your research goals and objectives.
Understand your data and codebook before asking AI Chat for suggestions.
Use filters to focus AI Chat on relevant codes or transcripts.
Always double-check AI outputs, especially snippet suggestions.
Start a new chat for a different topic to keep your analysis organized.
Common Example Prompts
Purpose | Example Prompt |
Summarize a code | "Summarize these snippets related to [code name]" |
Generate subcodes | "Help me categorize each snippet into more nuanced sub-codes with names, descriptions, and snippet citations" |
Provide examples | "Summarize snippets and include illustrative examples" |
Instructional guidance | "Write instructions for what type of snippet belongs in this code" |
Theory connection | "Situate this summary within [theoretical framework]" |
FAQs
Q: How do I view previous AI Chat conversations?
A: Click the three dots in the top right of the AI Chat panel, then select View All Chats. A list of past chats will open, and you can click to reopen any conversation.
Q: Can I delete a past chat?
A: Yes, click the trash icon to the far right of the chat title. Be careful—deleting a chat cannot be undone.
Want to learn end to end qualitative coding with AI?
Watch our 30 minute webinar where we break down the concepts of qualitative coding and how AI can be used to assist your process.