[homework] Reading & experience report

[siimre]

Asking

• [x] Ask ChatGPT for help completing this homework

Feeling, Writing, Thinking

• [x] Post an experience report (a paragraph or several) as a comment on this issue. This experience report should convey what motivates you to understand and practice data thinking. An experience report may describe a collection of experiences (which may or may not relate to data), and some questions to consider may be helpful: What were you or those involved feeling? Doing? Thinking? Asking? Where were they? What systems were involved, if any? Who would you want to better understand the incentives at play?
  • if you are stuck, ask for help on Zulip, come to office hours, make a plan of tasks to do and ask for feedback on the plan
  • try using dictation on iPhone, Android, computer, and ask ChatGPT to summarize it
  • ask Lex for ideas or ChatGPT for ideas
  • edit using https://hemingwayapp.com/, Grammarly, or any other tool to help
  • read the below articles for ideas

Doing

• [x] Review the first class: https://panopto.ut.ee/Panopto/Pages/Sessions/List.aspx?folderID=43bb180c-79a6-4324-b055-afa400ecd1a0 - ask on Zulip if you can't access this
• [x] Review the Figma whiteboard from the first class: https://www.figma.com/file/SsoNLhaZwnSGkADd4eS0D8/230220-introduction?node-id=0%3A1&t=jR2hISrDchAPapYR-1
• [x] Create a LinkedIn profile. (If you are not comfortable using your real name, feel free to use a fake name + email)
• [x] Post your LinkedIn profile on Zulip -- Zulip is open source and enables us to analyze our chat messages with natural language processing tools such as ChatGPT, which we will do later in the course!
• [x] Read https://matt.might.net/articles/how-to-email/ to understand principles of communication over email or Zulip
• [x] Read parts of the Mailchimp Style Guide (https://styleguide.mailchimp.com/voice-and-tone/ or https://styleguide.mailchimp.com/writing-principles/) to understand how written communication can be guided by principles
• [x] Read https://georgesaunders.substack.com/p/on-worry to understand a professional writer's perspective on writing.
• [x] Read https://every.to/chain-of-thought/writing-essays-with-ai-a-guide to understand how to use voice recording, dictation on your smartphone or computer, and ChatGPT, and Lex to help you write
• [x] Read https://simonwillison.net/2022/Dec/5/rust-chatgpt-copilot/ to understand how ChatGPT and CoPilot can help you learn data thinking, programming, and statistics skills
• [x] Read http://web.archive.org/web/20170208134519/http://storyshots.tumblr.com/post/25032057278/22-storybasics-ive-picked-up-in-my-time-at-pixar -- especially the "Once upon a time" story telling framework Pixar used for its movies.
• [x] Read https://www.newyorker.com/magazine/2023/02/13/the-dubious-rise-of-impostor-syndrome to help understand concepts that can increase our psychological safety and trust in class and how asking ChatGPT can help reduce our fear of asking for help when stuck.
• [x] Read https://jaan.io/my-friend-radicalized-this-made-me-rethink-how-i-build-AI/ to learn about my journey in understanding the incentives that can affect our thoughts, feelings, and actions in society and the incentives that affect how AI such as ChatGPT is built.

Programming & Infrastructure

• [x] Create a GitHub Codespace using the data thinking template here (from https://github.com/onefact/datathinking.org-codespace)
• [x] Send a screenshot to Zulip of a plot you made in the GitHub codespace (of any dataset or data you generated as in the live coding demo from the first lecture; ask for help if this is unclear)
• [x] Sign up for Lex - message Jaan on Zulip for an invite code
• [x] Sign up for Figma as a student (https://www.figma.com/education/students/)
• [x] Install copyq on Windows Powershell or Linux using Chocolatey, or install Maccy on Mac using Homebrew. This will save you a lot of time throughout the class as you copy and paste things from ChatGPT!

[siimre]

Writing tests with Copilot (https://til.simonwillison.net/gpt3/writing-test-with-copilot) was a very excellent article! It is under the "AI assisted learning: Learning Rust with ChatGPT, Copilot and Advent of Code" article (https://simonwillison.net/2022/Dec/5/rust-chatgpt-copilot/). It helps to save time and avoid routine.

I had ChatGPT search for information about Moonsund regattas from 2000 to 2022 (http://muhuvain.ee/tulemused-1/). I had it create a table showing the number of participating sailboats and the year of participation, and I also had it add a column for the number of sailors. Searching for these numbers one by one on the website would be quite a time-consuming task. I instructed ChatGPT to write Python code that generates a comprehensive graph, and it did so successfully. However, one major drawback is that the data is only sometimes accurate. Fortunately, ChatGPT can be trained by pointing out errors, and it will correct them.

Nevertheless, one should only sometimes believe everything ChatGPT responds with. For example, I asked some verification questions, such as who won in the ORC I class in 2019. It answered incorrectly. I am knowledgeable about this area, so I corrected it. I asked about the second place, and it responded poorly again - so mistakenly that the sailboat offered was not even in the ORC I class, and the skipper was not even from that sailboat! Fortunately, the trends depicted on the graph are correct. The number of participants was significantly lower during the economic downturn in 2008 and the year of COVID in 2020.