Historically, to answer questions about a text, you first have to read the text, and then attempt to answer the question. With modern AI tools, now you can do the opposite: get the answer first, then read the text as needed.
Think back to a recent time when you were assigned a reading and given comprehension questions. How did you approach it? Like many Brown students who work smarter and not harder, you might have skimmed the PDF, glanced at the questions, and then pasted them straight into an LLM along with the PDF. The LLM produced an answer, which you then checked / edited / verified by selectively returning to the reading.
We’ll call this process inverse reading: a strategy where you begin with a candidate answer generated by an LLM, and then talk to the LLM + read the text as needed to justify, refine, or correct that answer. The goal of this assignment is to think creatively about the emerging skill of inverse reading. You will be asked to do an inverse read of a text, and then you will design an augmentation to support inverse reading.
In Part A, you will practice inverse reading on one of the resources used to develop the lectures, specifically Chapter 9 from *Psychology of Reading* (2nd ed.) by Keith Rayner, Alexander Pollatsek, Jane Ashby, and Charles Clifton Jr.
Your job for Part A is structured into three tasks: 1) Interacting with the LLM. 2) Evaluating and repairing its response. 3) Reflecting on the entire process.
Each manual action you should carry out will be highlighted in a yellow callout box, like this:
<aside>
Setup: Create a document to write down your responses to this assignment.
</aside>
Each response you should provide an answer to in the document will be highlighted in a blue callout box, like this:
<aside>
Response: Write the header “CS1377 Assignment 2: Inverse Reading” at the top of the document. Don’t write your name!
</aside>
Preface before you begin:
<aside>
Setup: As you carry out Task 1 & 2, pay attention to how you are approaching the process. You will be reflecting on the entire experience in Task 3.
</aside>
First, you will start by getting an initial output from the LLM.
<aside>
Task 1.1: Download the chapter PDF by clicking the link below:
</aside>
<aside>
Task 1.2: Open up an LLM chatbot. Brown provides free access to Gemini, but you can use another one if you prefer. Set the LLM into its “smartest” mode, such as Pro for Gemini.
</aside>
Below are two comprehension questions that you will need to answer. Don’t try to answer them straight away, that’s just normal reading! Go look at the next task.