Artificial Intelligence (AI) continues to grow in popularity as it proves to be a useful tool for assisting with everyday tasks in people’s professional and personal lives. Not quite sure how to word that email? AI can help. Need help with workflow? Ask AI. Need a weekly meal plan and grocery list? AI can make those too.
But AI certainly has its limitations, and it’s important to understand what these limitations are when deciding whether or not to use it. This rings true when considering the use of AI in research, especially AI in qualitative research analysis. Striking a balance between optimizing efficiency and quality can be tricky, so let’s talk about some of the things to consider when deciding if using AI in qualitative analysis is right for you.
First, it’s important to know what AI is. While there is not yet a universally accepted definition of AI, in general, it refers to the ability for a system to synthesize information, then learn from that information to complete the assigned task. In qualitative data analysis, this means utilizing external information and knowledge to identify insights and themes from your qualitative data.
What are the benefits of using AI in qualitative research?
Qualitative research is a resource-intensive effort, making AI an appealing assistant to help lighten this load. In a fraction of the time that it would take to manually analyze the data, AI can organize large volumes of data and identify common themes. Sounds appealing! However, these AI systems often lack the ability to effectively identify the nuance, complexity, and context of the data that supports a thorough and thoughtful qualitative analysis.
AI Qualitative Analysis Concerns
While AI certainly has its benefits for efficiency and use of resources, it also has drawbacks. To revisit the definition of AI – AI systems are first fed information to synthesize, which then helps them complete the task, in this case, analyzing the qualitative data. As a result, what information you feed into the system directly influences how the system analyzes the data. How does this impact manifest?
Loss of context and nuance.
Qualitative data, especially in public health, is closely tied to the social, cultural, and even political context that exists within the community being studied. When AI systems analyze this data, this context can be missed, as AI systems lack the human element that allows for the consideration of these factors. Additionally, AI’s lack of understanding of the community can flatten some of the findings and fail to identify systemic connections between topics, resulting in a less robust qualitative analysis.
Bias.
As previously mentioned, developers train AI systems on specific information, which can include biases and skew toward Western norms, including the exclusion or underrepresentation of certain groups. This can result in analyses that echo these biases. While bias can also emerge in human-driven analyses, qualitative researchers utilize reflexivity to identify opportunities for bias and mitigate these impacts, a practice that is lacking in many AI-driven qualitative research analyses.
Risk of inaccurate results.
Using AI also comes with inherent risks. One of the more significant ones is the risk of hallucinations. AI hallucinations occur when the AI system generates an outcome that is misleading or even completely false. When considering this with AI in qualitative research analysis, this could include identifying a theme that does not reflect the data. This highlights another concern, which is the lack of transparency that AI-driven qualitative analysis provides. Human-driven qualitative analysis allows for an understanding of not only what the data is showing, but also how themes, quotes, and insights are identified.
Final Thoughts
So, what does this mean for you? Understanding the aims of your qualitative analysis and the resources available to you is an essential first step. For those with limited resources who are looking for quick, high-level qualitative insights, AI in qualitative research analysis may be a good option. However, for a comprehensive, nuanced, and high-quality analysis, human-driven qualitative analysis with trained researchers will provide insights that you can confidently utilize to understand the concerns and needs within your community.
Looking for a more thoughtful approach to qualitative research?
Crescendo supports population health and market research through interviews, focus groups, and more. Contact us to learn more.