# Site

## Pages

- [Page not found](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/404/): That page doesn't exist, or it has moved since you last looked
- [Assessment](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/assessments/): Every piece of graded work in the course, what it is worth, and when it is due.
- [Assignment 1: Build a predictive input system](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/assessments/assignment-1/): A working predictive-text build in a domain of your choice, submitted with a short written account of exactly where and why it fails
- [Assignment 2: Audit a deployed suggestion feature](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/assessments/assignment-2/): A comparative audit of a real, currently deployed prediction or suggestion feature, using the fixed-input method practised across weeks 6 through 9
- [Final project: The full audit](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/assessments/final-project/): The capstone — take Assignment 2's audit and turn it into a complete, standalone case study of a real product's suggestion design
- [Lectures](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/): What is covered, week by week, and which topics each lecture draws on.
- [Multi-tap, T9, and the keypad that started it](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-01/): The course's opening argument — that predictive text is a design lineage, not a recent invention, starting with the twelve-key keypad
- [Autocorrect and the comedy of failure](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-02/): Why the same system that quietly fixes most of your typos occasionally produces something you'd never have written on purpose
- [Query completion and the autocomplete confession](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-03/): What a search box's suggestions reveal about the people typing into it, and why that turned out to be a design problem
- [Prediction as access — AAC and assistive input](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-04/): What predictive text looks like when it's the only way someone composes language at conversational speed, not a convenience layered on top of typing
- [Code completion, from IntelliSense to Copilot](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-05/): Predictive text applied to a language with a grammar checker built in, and what changed when the completion stopped being confined to your own file
- [Smart replies and the suggested response](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-06/): What it means for a system to write the polite part of a conversation for you, one tap at a time, and why that started to feel strange
- [Conversational completion — the chatbot as autocomplete](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-07/): The first lecture back after the break, arguing that a conversational model is best understood as autocomplete operating at paragraph scale
- [Inpainting — completion in pixels](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-08/): What autocomplete looks like when the missing continuation is a region of an image instead of the next word, and why the two turn out to be related
- [Forecast UIs and the uncertainty they hide](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-09/): What a weather app, a poll tracker, and a stock chart have in common with autocomplete — and what each one does with the uncertainty a good forecast can't remove
- [Co-writing — authorship with a predictive collaborator](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-10/): What happens to the idea of authorship once the sentence you're writing is being continued, word by word, by something that has read more than you have
- [The feed — recommendation as prediction](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-11/): Why a recommendation feed is autocomplete for your attention rather than your sentence, and why that reframing changes what counts as a good guess
- [Persuasion by default — closing synthesis](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/lectures/week-12/): A guest lecture on consent by autocomplete, closing the semester by reading every prior week as a question about whose interests a suggestion serves
- [People](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/people/): Who teaches the course, and how to reach them.
- [Callum Voss](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/people/callum-voss/): Tutor. Ten years writing code-completion tooling before deciding the interesting part was never the ranking model.
- [Freya Bishop](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/people/freya-bishop/): Guest lecturer, week 12. A design researcher who studies how suggestion interfaces get people to say yes.
- [Priya Nandakumar](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/people/priya-nandakumar/): Founding convenor. Spent six years shipping predictive keyboards before returning to study why they work the way they do.
- [Policies and support](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/policies/): Late work, academic integrity, how predictive and AI tools may be used in this course specifically, and where to get help.
- [Build a T9 disambiguator](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/01-t9-and-keypads/): The first lab — turn a keypress sequence into a ranked list of real words, using nothing but a dictionary and a mapping of letters to keys
- [Autocorrect post-mortem](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/02-autocorrect-post-mortem/): A lab spent deliberately breaking an autocorrect implementation, then writing up why each break happened
- [Query completion audit](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/03-query-completion-audit/): A lab auditing a real search box's autocomplete for suggestions its owner probably wouldn't choose to put there on purpose
- [AAC scanning simulation](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/04-aac-scanning-lab/): A lab spent composing sentences through a scanning-and-selection interface with a deliberately mediocre word predictor, to feel the cost of a bad list
- [Assignment 1 clinic](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/05-assignment-1-clinic/): A work session for Assignment 1, run as one-on-one check-ins rather than a taught lab, in the week the assignment is due
- [Inbox smart-reply audit](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/06-inbox-smart-reply-audit/): A lab auditing your own email or messaging app's suggested replies against the messages that triggered them
- [Continuation-swap lab](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/07-continuation-swap-lab/): A lab where pairs swap half-finished conversations and try to tell a model's continuation from a human one, then check their reasoning
- [Inpainting seams lab](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/08-inpainting-seams-lab/): A lab spent generating and then deliberately hunting for the seams in inpainted images, cataloguing which tells actually generalise
- [Forecast UI comparison](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/09-forecast-ui-comparison/): A lab comparing how three real forecast interfaces represent uncertainty for the same underlying kind of prediction
- [Co-writing exchange](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/10-cowriting-exchange/): A lab where you write a short piece with a predictive tool, log every suggestion you accepted or rejected, and trade logs with a partner
- [Feed reverse-engineering](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/11-feed-reverse-engineering/): A lab where you deliberately engage with a narrow slice of content on a fresh or logged-out feed and log how quickly the recommendations shift
- [Closing crit](https://comp4020-agentic-coding-studio.github.io/comp4020-ass2-lilth2/sessions/12-closing-crit/): The last lab of the semester — short presentations of final project work in progress, judged against the persuasion framework from this week's lecture