Best practices for using AI features in your qualitative analysis

Who this article is for

This article is for researchers, students, instructors, and qualitative analysts who use AI-powered features in ATLAS.ti and want to apply them responsibly and effectively throughout their research process.

Why use AI in qualitative research?

Artificial intelligence can help researchers work more efficiently with large amounts of qualitative data.

ATLAS.ti AI features can assist with tasks such as:

• identifying concepts and patterns
• generating suggested codes
• summarizing data
• exploring themes
• supporting literature reviews
• accelerating exploratory analysis

AI can help researchers spend less time on repetitive tasks and more time interpreting findings and developing insights.

However, AI should support—not replace—the researcher's analytical judgment.

Understand the role of AI in qualitative analysis

AI is most effective when used as a tool to support qualitative analysis rather than as an automated decision-maker.

Think of AI as:

• a research assistant
• a source of suggestions
• a tool for exploring data
• a way to generate initial ideas

AI should not be viewed as:

• a substitute for coding decisions
• a replacement for methodological expertise
• a replacement for researcher interpretation
• a source of definitive conclusions

Qualitative research relies on human understanding, context, and interpretation, which remain essential regardless of how AI is used.

Use AI for exploration before confirmation

One of the most effective ways to use AI is during the exploratory phase of analysis.

AI can help researchers:

• identify recurring concepts
• suggest initial codes
• summarize large volumes of text
• surface potentially relevant quotations
• highlight patterns that may warrant further investigation

These outputs can help researchers become familiar with large datasets and identify areas that deserve closer attention.

However, AI-generated findings should be viewed as preliminary observations rather than confirmed results. Researchers should review the original data, validate AI-generated suggestions, and determine which findings are meaningful within the context of their research questions.

Use AI as a starting point

AI-generated outputs can provide a useful foundation for further analysis.

For example, AI-generated codes can:

• suggest potential themes
• identify recurring concepts
• highlight relevant passages
• support initial coding efforts

Researchers should then:

• review the suggestions
• refine the codes
• merge overlapping concepts
• remove irrelevant results
• develop their own analytical framework

The final coding structure should reflect the researcher's interpretation rather than relying solely on AI-generated outputs.

Verify AI-generated results

AI-generated outputs should always be reviewed and validated.

Before accepting AI suggestions:

• compare results against the original data
• verify quotations and supporting evidence
• review summaries for accuracy
• check whether suggested codes reflect participant meaning
• confirm that important context has not been overlooked

AI can occasionally produce incomplete, inaccurate, or misleading outputs. Verification helps ensure analytical rigor and accuracy.

Consider context during analysis

Qualitative research often relies heavily on context.

For example:

• cultural background
• participant experiences
• organizational settings
• emotional nuance
• sarcasm or humor
• implicit meaning

AI may not always recognize these contextual factors.

Researchers should carefully review AI-generated outputs to ensure that the original meaning of the data is preserved.

Be aware of bias

AI systems learn from training data and may reflect patterns or biases present in that data.

Potential issues include:

• overemphasizing common themes
• overlooking minority perspectives
• reinforcing existing assumptions
• generating biased interpretations

Researchers should actively evaluate AI-generated results and ensure that all participant voices are represented fairly.

Combine AI with established qualitative methods

AI works best when integrated into a well-defined research methodology.

Researchers can combine AI features with approaches such as:

• thematic analysis
• grounded theory
• content analysis
• framework analysis
• discourse analysis
• mixed methods research

The research method should guide the analysis, while AI supports specific analytical tasks.

Document how AI was used

Transparency is an important part of rigorous research.

Consider documenting:

• which AI features were used
• when AI was used during the project
• how AI outputs were reviewed
• how AI-generated suggestions were modified
• which decisions were made by researchers

This documentation can improve transparency, reproducibility, and methodological reporting.

Use AI responsibly with sensitive data

Researchers should always consider ethical responsibilities when working with participant data.

Before using AI-assisted workflows:

• review institutional requirements
• follow data protection policies
• consider confidentiality obligations
• evaluate ethical implications
• obtain required permissions when necessary

AI-assisted analysis should support ethical research practices rather than compromise them.

Combine AI insights with human interpretation

AI can help identify patterns, but interpretation remains the responsibility of the researcher.

Researchers should:

• evaluate the significance of findings
• determine relationships between themes
• develop theoretical explanations
• draw conclusions
• connect findings to research questions

The strongest qualitative analyses combine AI-assisted efficiency with human expertise and critical thinking.

As a guiding principle:

AI can inform, but human insight must guide and decide.

Recommended workflow for AI-assisted qualitative analysis

A practical workflow may look like this:

1. Import and prepare your data.
2. Review the data manually to become familiar with the content.
3. Use AI features to generate initial suggestions and exploratory insights.
4. Review and validate AI-generated outputs.
5. Refine codes and themes manually.
6. Compare findings against the original data.
7. Conduct deeper analysis using your chosen methodology.
8. Document how AI contributed to the analysis.
9. Generate reports and communicate findings.

This approach combines the efficiency of AI with the rigor of qualitative research.

Common issues and mistakes

• Treating AI outputs as final results
  • AI suggestions should be reviewed, refined, and validated before they are incorporated into the final analysis.
• Overlooking context
  • AI may not fully understand participant experiences, cultural context, sarcasm, humor, or implied meanings.
• Ignoring potential bias
  • Researchers should evaluate AI-generated outputs critically and ensure that findings accurately represent the data.
• Skipping manual review
  • Human oversight remains essential throughout the analysis process.
• Relying on AI for interpretation
  • AI can help identify patterns, but researchers remain responsible for developing explanations, themes, and conclusions.
• Not documenting AI usage
  • Recording how AI was used can improve transparency and support methodological rigor.

When to contact support

Contact ATLAS.ti Support if:

• AI features do not function as expected
• AI-generated outputs appear incomplete or inconsistent
• you encounter technical errors when using AI tools
• you need guidance on how a specific AI feature works
• you are unsure whether a behavior is expected or a software issue

When contacting support, include:

• your platform (Web, Windows, or Mac)
• your ATLAS.ti version if using Desktop
• screenshots or error messages
• the AI feature being used
• a description of the expected and actual behavior