search

onefact / datathinking.org

Data Thinking website deployed using GitHub Pages
https://datathinking.org
Apache License 2.0
7 stars 7 forks source link

[homework: asking, writing, thinking, doing, 🔴 red-teaming 😈] Critique what you have built; Real-World Data; Historiography of Data, Incentives, & AI #154

Open nshut opened 1 year ago

nshut commented 1 year ago

Reading

Pro tip: try using an app on your phone or computer to read aloud to you at 1.5x speed! This can save time and make it easier to absorb information while not being tied down to a computer or device visually.

Doing

Creating

Thinking

Listening

Large Language Model Access Checklist

nshut commented 1 year ago

My own critique:

I opened the report and am going to write whatever comes to mind during the next 10 minutes. Afterward, I will give the text I wrote to ChatGPT to make it more readable and coherent. As for me, the report itself looks nice. I feel that it looks "scientific." This is due to the template, but still. The word "Coauthor" should be eliminated because people don't explicitly call coauthors that, but this has a positive effect as well—it creates an impression of ChatGPT's agency. The first plot (Figure 2) does not have a title. The second plot (Figure 3) is impossible to read. It also has a red line in its legend, but I don't see it plotted. Figure 1 is a screenshot of a table, so one cannot copy anything from it, and the font is different from the main font used in the document. If I had more time, it would be better to recreate this table in LaTeX. Also, there are things written in italics and bold, but it is not intuitive what convention, if any, was adopted. Many technical aspects are not explained, making the report newcomer-unfriendly (I am new to this topic and don't understand everything that I did). Perhaps asking ChatGPT for an explanation of these things could help. References are listed as bullet points, which could be easily eliminated. Overall, the report looks like a "Potemkin village"—it looks smart but doesn't have much substance behind the numbers and plots.

ChatGPT critique

Thank you for sharing your homework. After reviewing the content, here are my critiques:

  1. Writing:

The writing is generally clear and well-structured. You have a good introduction, clear section headings, and a conclusion, which makes the paper easy to follow.

There are a few places where the language could be more formal or academic. For example, phrases like "I tried to start doing this homework several times" could be rephrased to something like "Initial attempts at this assignment were made using various IDEs..."

There are a few typographical errors, such as "Tis time" which should be "This time". Proofreading the document carefully before submission would help catch these.

  1. Visualization:

Without seeing the actual visualizations, it's hard to critique them. However, based on your descriptions, it seems like you've used appropriate methods for visualizing your data.

Make sure your visualizations are clearly labeled and include a legend where necessary. This will make it easier for readers to understand what they're looking at.

  1. Content:

The content is generally good. You've clearly explained your methods and the results of your analyses.

However, the results section could be improved by providing more interpretation of the results. For example, you mention that the accuracy of the logistic regression model was 0.3125, but you don't discuss whether this is a good result or not, or what it might mean in the context of your data.

Similarly, you mention that the performance of the linear regression model was not very good, but don't provide any details about why this might be the case or how it could be improved.

  1. Improvements:

The paper could benefit from a more detailed discussion of the results and their implications. This would make the paper more interesting and valuable to readers.

Including the actual visualizations in the paper would also be a significant improvement. This would allow readers to see the results of your analyses for themselves.

Finally, proofreading the paper carefully before submission would help catch typographical errors and improve the overall quality of the writing.

I hope these critiques are helpful! Let me know if you have any questions or if there's anything else you'd like me to review.

nshut commented 1 year ago

In my project, I will aim to explore the concept of "fig leaves" - utterances or actions that provide cover for statements or behaviors that would otherwise be seen as racist - in the context of Donald Trump's tweets (Notable example of a fig leave is "I am not a racist, but very racist claim"). I will try to create a search engine capable of finding tweets relevant to a given topic, thereby enabling the analysis and understanding of the use of fig leaves in Trump's discourse on Twitter. The project is driven by the challenge of applying machine learning and natural language processing techniques to real-world, complex dataset, and a desire to shed light on the subtle ways in which language can be used to mask or deflect from problematic statements or actions.