Date: Thursday, November 7, 2024
Hi Everyone, my name is Najat Elgeberi and I work as an assistant professor for program evaluation at the University of Nevada Reno-Extension. In my career I used to do a lot of consultancy work in the field of evaluation. I had to analyze data, create visuals and write reports. The most challenging part was how to create the charts that reflect data in an accurate way and at the same time are visually appealing. This task used to take a lot of time from me, as usually I can be a perfectionist.
When AI emerged, I started wondering to what extent can AI produce good looking charts that provide a good representation of data. My curiosity pushed me to try Chat GPT and see to what extent it can out beat me in creating charts from the same data I use. I provided a set of commends shown in the following thread of pictures and monitored what will this machine create.
First, I asked it to improve the chart that I created from my data (in a format of picture) to make a more visually appealing chart. The results are below:
Second step, I decided to upload the table that I used to create the chart as an Excel sheet to Chat GPT and provide the following command “Use the attached sheet to create a visually appealing bar chart”. The results are shown below:
The results were not satisfactory to me, so I asked AI to “Use principals of data visualization to improve the previous chart and transforming it to a better visually appealing chart”, So AI generated the next picture. And provided me with a list of key principals it used to generate the chart, AI said: “The updated chart now follows key data visualization principles:
My curiosity pushed me to ask AI to “Use the best type of charts that fit the type of data used in the excel spreadsheet attached previously”. I wanted to know if AI can provide the best judgement by itself on the type of chart that should be created and below was the result.
The results were very bad for me and disappointing, so I asked AI to improve the first chart it created using the data (not my original picture) and provided multiple commends for improvements till I got the below result.
After this short experiment I became convinced that AI is only smart when it relies on us or when provided with the right directions. When left to its best judgement it doesn’t always provide the best results especially with data visualization. The good think that I liked and I think I will use in the future is when I asked it to “Write a caption that explains the chart for people who are visually impaired”, The AI provided the following statement:
“This bar chart visually represents the top challenges faced by teachers, as gathered from a survey. The horizontal bars compare the percentage of respondents who selected each challenge. The chart is ordered with the most selected challenges at the top and the least at the bottom. Each bar also has a percentage label at its end, showing the exact proportion of respondents who mentioned the challenge.
For example, the longest bar at the top represents “Lack of professional development opportunities,” with around 9% of respondents identifying it as a key challenge. The bars are shaded with a gradient of blue, with darker shades representing higher percentages, helping to highlight the most significant challenges.”
I will leave the judgement to you to determine if the caption was really a good one or not and whether you can trust AI to give you results that out beat you ?
The American Evaluation Association is hosting Data Visualization and Reporting (DVR) Week with our colleagues in the DVR Topical Interest Group. The contributions all this week to AEA365 come from DVR TIG members. Do you have questions, concerns, kudos, or content to extend this AEA365 contribution? Please add them in the comments section for this post on the AEA365 webpage so that we may enrich our community of practice. Would you like to submit an AEA365 Tip? Please send a note of interest to AEA365@eval.org. AEA365 is sponsored by the American Evaluation Association and provides a Tip-a-Day by and for evaluators. The views and opinions expressed on the AEA365 blog are solely those of the original authors and other contributors. These views and opinions do not necessarily represent those of the American Evaluation Association, and/or any/all contributors to this site.