Summarizing data for continuous improvement - with a little help from AI

The tool: Qualitative Coding and Chat GPT

The purpose: To start qualitative coding, just a little bit faster 

Use case: 

If you read my last blog post, you are familiar with using qualitative coding to analyze data and measure impact. But what if you are having trouble uncovering the main themes or the data? Or maybe you just want to get a head start on coding your data, and you need a brainstorming partner. Meet the new kid on the block, Chat GPT. 

Instructions: 

1. Download and clean your survey data. Remember, garbage in = garbage out, and that is especially true when using AI. 

2. If you are doing this for work, make sure to remove all identifying information from your data, and check with your IT department regarding their policies around what information can and cannot be shared with generative AI

3. Open https://chat.openai.com/auth/login and login 

4. I like to start off with the prompt “Summarize the following lines of data and provide counts of each instance.” 

5. Then, copy-paste your data into Chat GPT. 

6. The platform will produce something like, “Here is a summarized overview with counts of each instance: 

1. No Specific Response:

1. "N/A"

2. "na"

3. "NIL"

4. Count: 3

2. Positive Feedback:

1. "It's an excellent module."

2. "More simulations can help."

3. "Increase different cases."

4. "So far so good."

5. "No comments."

6. Count: 5

7. Now, here’s where human, rather than artificial, intelligence comes into play. Is “more simulations can help” Actually a positive comment? Or is it a suggestion for improvement? Start making those calls. 

8. You can continue your prompts until you are happy with your summary. Make sure to validate your findings, rather than just turning in an evaluation summary created only by AI. 

The Pros: 

Used correctly, Chat GPT can save you time in developing the main themes in your data. This saves valuable time reading and grouping every comment. 

The Cons: 

Chat GPT is NOT going to tell you which recommendations are feasible for your time and budget. Its ability to apply best practices for adult learning is.. Marginal at best, but getting better every day. But for summarizing qualitative data quickly, it is useful. 

You may also need to consider your company's data privacy policies - is the information you are sharing with the platform from proprietary training? If so, proceed with caution. 

The Takeaways: 

I buy the hype that Chat GPT is here to stay, and in the next few years, AI will be built into almost every tool Instructional Designers use. As newer and better versions roll out, this post could be obsolete - but for now, this is one use case for this new tool!