AI100

AI100 LLM Continuation Guide

Use this document to continue the AI100 exam-preparation work with another LLM. Read it before changing or creating any study material.

1. Project Goal

This repository publishes concise, accurate, printable, and browser-friendly AI100 exam resources.

The user may provide new:

• textbook chapters;
• lecture PPT/PPTX files;
• study guides;
• multiple-choice review questions and solutions;
• theory questions and solutions.

For each exam, synthesize those sources into:

1. a two-sided A4 cheat sheet in standalone HTML; and
2. concept slide decks for topics the user finds difficult.

The materials must prioritize the official study guide and review/theory questions over generic textbook summaries.

2. Current Published Resources

• AI 100 Midterm I - Portrait Large-Print Cheat Sheet
• AI 100 Chapter 4 - McCulloch-Pitts Neurons and Finite Automata

Current repository:

• GitHub: https://github.com/ImDrCee/AI100
• GitHub Pages: https://imdrcee.github.io/AI100/
• Default branch: main

3. Existing File Roles

Published files

• AI100_Midterm_Two_Sided_Cheat_Sheet_Portrait.html
  • Two-page A4 portrait cheat sheet.
  • Uses two balanced columns per page.
  • Body text is approximately 8 pt.
  • Designed for duplex printing with long-edge flipping.
• AI100_Chapter4_Neurons_and_Finite_Automata_Slides.html
  • Standalone 25-slide HTML presentation.
  • Covers McCulloch-Pitts neurons, Boolean prerequisites, finite automata, recurrent memory, and the finite-automaton/Turing-machine boundary.
  • Includes relevant official Review and Theory question references.
  • Supports arrow keys, Space, Home, End, touch swipes, URL slide fragments, and A4 landscape printing.
• README.md
  • Contains direct links to the rendered GitHub Pages resources.

Local source files that must not be published automatically

The original working folder may contain a textbook draft, course slides, study guides, questions, and solutions. Treat these as source material for analysis, not as files to publish.

In particular:

• Never upload the textbook draft.
• Do not upload lecture PPT/PPTX files, review PDFs, solution PDFs, or other instructor-provided material unless the user explicitly requests it and confirms they have permission.
• Publish generated study aids only by default.

4. Source Priority

When sources disagree or space is limited, use this order:

1. Official exam study guide and stated exam scope
2. Official review questions and solutions
3. Official theory questions and solutions
4. Lecture PPT/PPTX material
5. Assigned textbook chapters
6. General background knowledge

Preserve the course’s terminology and conventions, even when another field or textbook commonly uses a different term. If there is a meaningful ambiguity, state the course convention and briefly note the alternative.

5. Workflow for the Next Exam

Step 1: Confirm scope

Ask the user for:

• exam name or number;
• included chapters;
• included lectures;
• whether all PPTs are in scope;
• permitted cheat-sheet page count and orientation;
• whether answers may appear directly on concept slides;
• whether existing files should be preserved as previous-exam versions.

Do not overwrite a previous exam’s published file. Create a clearly named new file, for example:

AI100_Midterm_II_Two_Sided_Cheat_Sheet.html
AI100_Final_Two_Sided_Cheat_Sheet.html
AI100_Chapter7_Concept_Slides.html

Step 2: Inventory sources

List all relevant files and identify:

• book chapter boundaries;
• lecture dates and topics;
• study-guide sections;
• every review question in scope;
• every theory question in scope;
• corresponding official solutions.

Do not assume PDF text extraction preserved diagrams correctly. Render and inspect pages containing state diagrams, graphs, tables, equations, or other visual questions.

Step 3: Build a question-to-concept map

Before authoring, map each question to:

• concept tested;
• required definition or procedure;
• official answer;
• likely distractor or common mistake;
• lecture/chapter source.

Every in-scope review and theory question must appear in the final map. Place question-reference badges beside the relevant concept, not only in an appendix.

Step 4: Write the cheat sheet

The cheat sheet should emphasize:

• exact definitions;
• formulas and algorithms;
• compact comparison tables;
• worked calculation patterns;
• state-tracing procedures;
• official answer cues;
• common traps;
• short theory-answer templates.

Prefer readable compression over tiny text. Use the available page area efficiently and balance columns.

Step 5: Write concept slides

Concept slides should:

• begin with an intuitive mental model;
• introduce formal notation only after the intuition;
• show a repeatable solving procedure;
• include worked examples;
• explain why wrong approaches fail;
• connect related computational models;
• display relevant Review and Theory question badges;
• end with a complete question-reference map and mastery checklist.

The slides must be a standalone HTML file with no external runtime dependency.

Step 6: Validate

For cheat sheets:

• emulate print media;
• verify the exact requested A4 page count;
• verify portrait/landscape dimensions;
• check every column and element for overflow or clipping;
• verify text remains searchable;
• visually inspect both/all printed pages.

For slides:

• test at a 16:9 viewport such as 1600x900;
• ensure every slide has no horizontal or vertical overflow;
• verify keyboard navigation;
• verify URL fragments such as #14;
• verify there are no browser console errors;
• verify print output produces one A4 landscape page per slide;
• visually inspect representative title, formula, diagram, table, and final reference slides.

Do not report completion until these checks pass.

6. HTML Design Requirements

Cheat sheets

• Use @page with explicit A4 size and orientation.
• Use physical units (mm) for page dimensions.
• Use break-after: page or page-break-after: always.
• Enable print-color-adjust: exact.
• Hide browser-only controls during printing.
• Avoid content that depends on an internet connection.

Slides

• Use one .slide section per slide.
• Provide visible previous/next controls and a slide counter.
• Support:
  • Right/Down/PageDown/Space for next;
  • Left/Up/PageUp for previous;
  • Home and End;
  • touch swipes;
  • URL hashes for direct slide access.
• Use inline SVG for automata, graphs, and diagrams.
• Keep formulas and answer cues large enough to read from a screen.
• Add print CSS for A4 landscape.

7. Current Course Conventions and Known Ambiguities

Retain these conventions if the topics reappear:

• McCulloch-Pitts output is 1 when sum(weight_i * input_i) >= threshold; otherwise it is 0.
• AND: weights (1,1), threshold 2.
• OR: weights (1,1), threshold 1.
• NOT: the course slides/practice solutions emphasize weight -1 and threshold -0.5.
• The textbook may show NOT with threshold 0; both work for binary input under the stated >= rule. Prefer the repeated course convention and note the alternative only when useful.
• A single threshold neuron cannot compute XOR/XNOR because they are not linearly separable.
• A finite automaton accepts only when it ends in an accept state after the entire finite input has been read.
• An accept state is drawn with a double/concentric circle.
• The language of an automaton is the set of all strings it accepts.
• A complete DFA for exactly 101 needs a rejecting sink and transitions for every wrong or additional symbol.
• The course summary is: finite automaton + external read/write memory = Turing machine.
• Course slides may use “admissibility” to mean guaranteed optimality in search. Standard later AI terminology often uses “optimality” for the property and “admissible” for a non-overestimating heuristic.

8. Git and Publishing Rules

• Repository: ImDrCee/AI100
• Keep the repository private unless the user explicitly changes that choice.
• GitHub Pages output is public.
• Upload generated HTML study resources and documentation only by default.
• Never upload the textbook or other instructor-provided source files without explicit permission.
• Never overwrite an existing published exam resource unless explicitly asked.
• Use descriptive commit messages, for example:

Add Midterm II printable cheat sheet
Add Chapter 7 concept slides
Update study resource links

• After publishing, verify the exact github.io URL returns HTTP 200 and renders as a webpage rather than source code.
• Add every new public resource to README.md using its GitHub Pages URL.

9. README Link Format

Use ordinary Markdown links:

[Resource title](https://imdrcee.github.io/AI100/FILENAME.html)

GitHub Markdown cannot force links to open in a new tab. Users can use Ctrl+click, Cmd+click, or the browser’s “Open link in new tab” command.

10. Completion Checklist for the Next LLM

Before concluding a future exam update, confirm:

• Exam scope was confirmed.
• All source files were inventoried.
• Every in-scope Review question was mapped.
• Every in-scope Theory question was mapped.
• Official solutions were checked.
• Diagram-based questions were visually inspected.
• Existing published files were preserved.
• New HTML passes screen and print validation.
• Only approved generated files were uploaded.
• GitHub Pages URLs were verified.
• README.md was updated with all new links.
• The user received direct links and print/navigation instructions.