Leveraging AI Tools for Qualitative Data Analysis in Social Impact Research: A Review of ATLAS.ti

In today’s technology-driven society, Artificial Intelligence (AI) is revolutionizing multiple aspects of our daily lives. From automating routine and time-consuming tasks to adding rigour and consistency to scientific methods, the transformative potential of AI is undeniable. Notably, the field of research and evaluation has become acquainted with these advancements. Traditionally, AI in research has primarily been recognized as a tool to handle large quantitative datasets. However, recent adaptations have led to the realization that the capabilities of AI can also extend to the domain of qualitative research, and support sifting through vast sets of interview transcripts. ATLAS.ti is a pioneer software for this type of qualitative data analysis.

ATLAS.ti is a cutting-edge software designed to harness the power of AI in qualitative research, using OpenAI’s GPT model [1]. The primary objective of this software is to support researchers in extracting insights from textual data, making the analysis process more efficient. The software also offers visualization features for more comprehensive analysis and summarization of findings.

Having trialled this software with a set of transcripts I had previously analyzed, I have formed some initial impressions of the value of this software, in its current version, to support seamless qualitative analysis in the social research field.

Key Advantages

The AI summary in ATLAS.ti is able to instantly provide a general summary of each transcript. Researchers can quickly understand the core topics and ideas discussed in an interview which saves the time it takes to meticulously read through the entire transcript.

Significantly, an autocoding feature can substantially reduce the time and effort required by automatically producing codes for a set of transcripts. This is an invaluable tool for researchers when handling large datasets. The initial set of codes can then be modified, grouped, and refined. If the researcher prefers to review the transcript themselves, then ATLAS.ti can provide smart coding suggestions. Potential codes can be highlighted whenever the user highlights a snippet of text. This feature can enhance the productivity ofthe coding process while allowing the user to make the final decision, supporting and building a robust coding framework.

Sentiment Analysis is a feature that researchers can use to categorize data based on its emotional tone, further simplifying the process of gleaning insights by indicating the overall mood of a section of text.  With visualization capabilities ATLAS.ti can offer multiple visualization tools, such as word frequencies and diverse rapid visualization formats including tree maps or donut diagrams. These tools can be used to provide a holistic view of the coded data, facilitating a deeper understanding.

Drawbacks and Limitations

While AI is powerful, it cannot always grasp nuanced human emotions and expressions. Sentiment analysis may falter when confronted with certain tones or sarcasm that only a human perspective can truly interpret. There is often a conflict between consistency and context. While AI can offer consistent analysis and improve accuracy in terms of avoiding human cognitive fatigue, it may sometimes miss the context that a human researcher can appreciate more fully.

Autocoding can be limited. ATLAS.ti’s autocoding only offers an inductive approach with code emerging from a body of text already provided. There is an unmet need for a more deductive feature, where a predefined set of codes can be systematically applied to transcripts. This could greatly save time for researchers while maintaining their initial coding framework that has been specifically tailored to the objectives of the research.


Implementing AI tools such as ATLAS.ti, presents a promising outlook for revolutionizing qualitative research processes. In its current state, it can provide unparalleled efficiency. However, researchers must employ such tools conscientiously, complementing its strengths with human intuition and understanding. With the right balance between AI tools and human input, future technological adaptations can expect to redefine the landscape of social impact research.

GtD is your expert partner to deliver quantitative analysis on your project. We will use qualitative analysis to gain deeper understanding. GtD are always looking to adopt the latest technology to deliver analysis more efficiently and economically. We look forward to hearing from you with all your data needs.


[1] OpenAI is one of the leading research organizations in AI. Their GPT models have been trained to comprehend natural language and code. Hence, ATLAS.ti have tailored OpenAI’s GPT model to create a system which automatically codes and summarizes bodies of text.