I have been experimenting more and more with ChatGPT on a regular basis. I’m interested and curious how this will help. I had a friend at one of the large FANG companies note that they found ChatGPT to be very helpful in their job, with assistance in various things. Their quote:
”…I used it to collaborate, having a design discussion, writing requirements, comments and document my code (!!), improve my writing and find the gap and provide suggestions…”
I thought that was interesting, so I sent in this prompt: write a one page post in the style of http://www.voiceofthedba.com that summarizes this article https://www.madeiradata.com/post/home-experiments-and-stacked-bar-graphs
I often find links that catch my eye, read them, and save them in a note for a future editorial. This time I asked ChatGPT.
What I got from ChatGPT is below. I read it and then wrote my own editorial, which was called The Best Graph. Mine is quite different, and I think, it’s better. I’m not sure the response below summarizes the post correctly, I don’t know it’s in my style (maybe parts), and ultimately, it’s a summary, not a commentary. Maybe that’s on me. I didn’t specify an editorial or opinion.
Read them both and let me know which one you like more, which you think works better.
The ChatGPT Response
In a recent blog post on Madeira Data Solutions, author Lisa B. shares her findings on the use of stacked bar graphs in visualizing data from home experiments. While stacked bar graphs are commonly used in business and marketing contexts, Lisa explores their potential usefulness in a personal setting.
Lisa begins by describing her home experiments, which involve testing the efficacy of various cleaning products. She notes that while she initially used line graphs to track the results of her tests, she found that stacked bar graphs were more effective in illustrating the relative performance of each product.
Lisa goes on to provide a detailed explanation of stacked bar graphs and their components, including the use of color-coding and labels to differentiate between different categories and subcategories. She also notes that stacked bar graphs are useful in highlighting changes over time and in comparing data across different groups.
One of the key takeaways from Lisa’s post is the importance of choosing the right type of graph for a given dataset. While line graphs may be appropriate for tracking changes in a single variable over time, stacked bar graphs may be more useful when comparing multiple variables or subcategories. By selecting the appropriate type of graph, researchers can more effectively communicate their findings and insights to others.
Overall, Lisa’s post provides a useful overview of stacked bar graphs and their potential applications in a personal, as well as professional, context. Whether you’re conducting home experiments or analyzing business data, it’s important to choose the right type of graph to effectively convey your findings.
Lisa B.? Who’s that? That article was written by none other than Rita Fainshtein! 😅
That’s crazy. didn’t even notice the wrong author
My name is Rita Fainstein, and I am the author of the article “Home Experiments and stacked bar graphs” on Madeiradata.
It is evident that both the GPT chat summary and the second version written by you differ greatly from the content of the article.
Every aspect of Chat GPT’s “product” was distorted, from my name to the weird stuff he made up about cleaning products. That is a very funny thing to see :).
There is no doubt that the summary you wrote is more intelligent, but it is also apparent from it that there was no concentrated reading of the article and consideration of its content.
Your comparison does not relate to the scenario given in the example I wrote.
As a data visualization specialist with over four years of experience, I am familiar with Edward Tufte and Stephen Few’s works. In most cases, there is only one graph that can be used to convey the message that pertains to the brain’s perception.
There is a graph for each scenario/message. A graph’s type is determined by the question it answers, rather than by the data set it represents
It is true that if my article had discussed changes over time, it would have discussed a line graph. However, the topic is quite different…
Also, it is a “great surprise” that no mention is made of histograms…
Conclusions regarding the GPT chat
The content of Chat GPT is currently beyond our control, but we have a responsibility as human beings to examine it thoroughly before making a judgment.
LikeLiked by 1 person
Thanks for the note. The response from the AI is definitely strange.
ChatGPT and similar bots are tools and nothing more. We all use intelisense in much of our development tools but we don’t always assume that intelisense is perfect . The same should be applied to ChatGPT . You use it to generate an overall framework that you then review and modify as needed.
that’s what I find. It’s helpful, lightly, but not a lot so far because of the editing requirements. I’ll keep working on my prompts and see if that changes.