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

[Chartiza]

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.

• [x] Review the bug report we submitted: https://github.com/duckdb/duckdb/issues/7015 and the fix to avoid the bug we ran into for analyzing Zulip data: https://github.com/onefact/datathinking.org-codespace/blob/main/notebooks/in-class-notebooks/230420-zulip-chat-duckdb-loading.ipynb
• [x] Review several editions of Style, Clarity and Grace (https://bookshop.org/p/books/style-lessons-in-clarity-and-grace-joseph-williams/8980182 - Libgen)
• [x] [climate harms] Read https://arxiv.org/ftp/arxiv/papers/2104/2104.10350.pdf
• [x] [global context thanks to @indrekromet] Read https://www.palladiummag.com/2023/02/23/the-west-lives-on-in-the-talibans-afghanistan/
• Preparation for guest lecture
  • [x] Read https://arxiv.org/abs/2302.10329
  • [x] https://www.economist.com/science-and-technology/2023/04/05/it-doesnt-take-much-to-make-machine-learning-algorithms-go-awry (https://archive.ph/5l1k3#selection-1038.0-1038.1)
• [x] [David Foster Wallace on Consumption] https://harpers.org/wp-content/uploads/2008/09/HarpersMagazine-1996-01-0007859.pdf
• [x] [Cross-Cultural Research and Ethnography] Read excerpt from https://www.thefieldstudyhandbook.com/ - https://drive.google.com/file/d/1F9zr1pHesGPSOTtXLQu4hTxOGu7zoIRx/view?usp=sharing
• [x] [Status games, social engineering] Read the chapter on status games in Impro (https://bookshop.org/p/books/impro-improvisation-and-the-theatre-keith-johnstone/11721204 or Libgen)
• [x] [Closure on the data journey begun in the first class on COVID] https://www.wsj.com/articles/covid-origin-china-lab-leak-807b7b0a
• [x] [History of Computing and Data Thinking] Read the Epilogue of Computer Age Statistical Inference by Hastie et al., https://hastie.su.domains/CASI_files/PDF/casi.pdf
• [x] [Basics of GitHub] https://github.com/onefact/github-starter-course
• Systems Thinking and Incentives:
  • [x] [Context on how big 4 consulting firms may/may not be replaced by AI] https://www.lrb.co.uk/the-paper/v44/n24/laleh-khalili/in-clover (https://archive.ph/yC1YQ)
  • [x] [High finance & geopolitics] https://www.theguardian.com/world/2016/jul/28/1mdb-inside-story-worlds-biggest-financial-scandal-malaysia
  • [x] [High finance & geopolitics] https://www.bloomberg.com/news/articles/2014-08-21/mugabes-bailout-och-ziff-investment-linked-to-zimbabwe-despot#xj4y7vzkg?leadSource=uverify%20wall
  • [x] [Predatory Investing] https://www.newyorker.com/magazine/2018/08/27/paul-singer-doomsday-investor
  • [x] [Global Geopolitics of AI] Chapter 1 - China's Sputnik Moment - and Chapter 4 - A Tale of Two Countries - of Kai-Fu Lee's AI Superpowers (https://bookshop.org/p/books/ai-superpowers-china-silicon-valley-and-the-new-world-order-kai-fu-lee/6960836 or Libgen)

Doing

• [x] Edit https://en.wikipedia.org/wiki/Data_thinking using the learner definitions of data thinking
• [x] Try playing one round of https://lost-at-sql.therobinlord.com/
• [x] Install https://marketplace.visualstudio.com/items?itemName=charliermarsh.ruff and read the documentation, and activate the extension for python files opened in Visual Studio Code: https://beta.ruff.rs/docs/tutorial/
• [x] Red-teaming Previous Homework:
  • [x] [Set timer: 10 minutes maximum] Before asking GPT (to avoid biasing yourself!), write your own critique of your homework. Questions to consider could be: what could be improved? What doesn’t make sense in the visualization? What doesn’t make sense in the writing?
  • [x] [Set timer: 10 minutes maximum] Ask GPT-4 to critique the homework or the visual using your favorite data thinking definition we have so far.
  • [x] [Set timer: 10 minutes maximum] Add this critique as a comment on the homework’s github issue, and link to the critique in Zulip.
  • [x] [Set timer: 10 minutes maximum] Repeat this exercise for the previous homework of one other person in the class

Creating

• [x] Generate art or diagrams or logos using https://beta.dreamstudio.ai/generate and post on Zulip chat
• [x] Use https://poly.cam/ to scan an object or location of your choosing; share the link on Zulip (bonus: feel free to try BlenderGPT: https://github.com/gd3kr/BlenderGPT to modify the 3D scan)
• [x] Use duckdb to load the Zulip data into a SQL database, and use altair to visualize the data, following https://github.com/onefact/datathinking.org-codespace/blob/main/notebooks/in-class-notebooks/230420-debugging-duckdb-altair-falcon-3-1-1-service-requests.ipynb (run this notebook with this data: https://data.cityofnewyork.us/Social-Services/311-Service-Requests-from-2010-to-Present/erm2-nwe9 - and try changing the data source to be the Zulip data and post a visualization of the chat data on Zulip)

Thinking

• [x] In a comment on this issue, propose a research question that you will aim to answer the upcoming final homework (and cross-post it on Zulip for feedback); include a rationale for why this topic matters to you, and decisions you need to make, and a clear articulation of the stakes. This can be a paragraph or several. Perhaps start with your experience report from the first homework.

Listening

• [x] Listen to this podcast with Emily Bender: https://open.spotify.com/episode/2ILGlkAXAt4xfuKHwIV2on

Large Language Model Access Checklist

• [x] Try GPT-4 at https://chat.openai.com/chat (ask on Zulip if you need access)
• [x] Try https://serge.chat/
• [x] Try https://bard.google.com
• [x] Try https://console.anthropic.com/claude (ask on Zulip if you have not received an invite)
• [x] Try https://www.phind.com/
• [x] Try Copilot (https://github.com/features/copilot; free with GitHub Education Pack)
• [x] Try Bing Chat (https://bing.com/chat)
• [x] Try Huggingface Chat (hf.co/chat)

[Chartiza]

Critique to my previous project. link to project pdf: datathinking_pantiukh.pdf

The main question I have is about visualizing linear and logistic regression predictions. Based on both figures, I see that prediction does not work. But the problem is in the model, not in the visualization technique. Moreover, a comparison of model-based prediction and random prediction would be useful, but it wouldn't make sense in this case due to the poor performance of the model.

[Chartiza]

GPT 3.5 critique to my previous project.

The visualization in your project demonstrates the use of t-SNE for dimensionality reduction and visualization of embeddings. The scatter plot shows the transformed embeddings in a two-dimensional space, with different colors indicating the assigned sender IDs. This approach allows for a visual understanding of the distribution and clustering of the embeddings based on the sender IDs.

One potential critique of the visualization is the lack of clear labeling or legend to indicate the correspondence between the colors and the specific sender IDs. This could make it difficult for viewers to interpret the plot accurately without additional context or guidance. Providing a legend or labeling the plot with corresponding sender IDs would enhance the clarity and understanding of the visualization.

Additionally, it would be beneficial to include a brief explanation or interpretation of the observed patterns or clusters in the scatter plot. This would help readers to grasp the insights or implications of the visualization and understand how the sender IDs relate to the distribution of embeddings.

Overall, while the t-SNE scatter plot provides a visual representation of the embeddings, some improvements, such as labeling and interpretation, would enhance the effectiveness and clarity of the visualization.

[Chartiza]

Other person project critique: link for the report: My_Working_datathinking.pdf

Overall, I liked this report. I did not review other people's reports while I was working on my own report and in this report I see plots that fit very well with my own. For instance, I like that the author gave us a learning curve for understanding the model and formulas. However, I would prefer more accurate picture placement. There are a lot of pictures at the end of this page, so it looks a bit cluttered. Also I would like to see some references at the manuscript.

[Chartiza]

final homework project:

LIFEWISE (Life Index for Well-being and Expense Sustainability) project utilizes regression analysis to identify optimal places to live, considering the balance between well-being and expenses. It offers actionable insights for individuals seeking high-quality living within their budget, while providing decision-making support for policymakers, urban planners, and real estate developers. The project aims to empower individuals and communities to make informed choices that enhance their overall quality of life and long-term sustainability.

The topic of LIFEWISE holds deep personal significance to me, as it is rooted in my own journey of seeking a new home in Europe. Having been forced to move because of war, I found myself faced with the daunting task of deciding where to rebuild my life. With limited knowledge about European countries and their living conditions, I yearned for a data-driven approach to guide my decision-making process. It was this pivotal moment that sparked my interest in exploring the balance between quality of life and cost of living. Through the LIFEWISE project, I aim to empower individuals like myself, who seek a fresh start in unfamiliar territories, by providing actionable insights that enable informed choices.