Human responsibilities
The overall concept, learning objectives, course fit, stage decomposition, grading logic, and the final decisions about what belongs in the assignment all remained human responsibilities.
COMP10002 Foundations of Algorithms
How This Assignment Was Built
A short note on why this assignment was developed as a human-AI collaboration, and why that improved the final result for students from Dr. Cohney's perspective.
Short answer
This assignment was built as a human-AI collaboration between Dr. Cohney, Dr. Cohney’s AI tool and Kacie Beckett.
The teaching goals, the algorithmic structure, the correctness requirements, the marking priorities, and the final editorial decisions are human. AI was used as a fast design and production partner during drafting, iteration, rewriting, visual explanation, and polish.
Why Do This At All?
The blunt version is: using AI let me build a fancier, cooler, more interesting, more realistic, and more pedagogically valuable assignment than I could have produced alone under the same time and energy constraints.
Why use AI here
The point was not to remove human thinking. The point was to increase the amount of thoughtful work that could actually fit into one assignment.
With AI helping, it became practical to build:
- a web-first spec rather than a single flat handout
- linked concept explainers for ideas such as softmax, masking, and attention
- alternate explanations aimed at different levels of prior knowledge
- more diagrams, examples, glossary support, and implementation guidance
- tighter iteration on wording, sequencing, structure, and visual design
That matters because a lot of assignment quality is not in the core algorithm alone. It is in the surrounding teaching infrastructure: how clearly the task is introduced, how well the ideas connect, how easy it is to recover when confused, and how much support exists.
Division of labour
AI responsibilities
AI helped generate alternatives, rewrite explanations, draft examples, propose page structures, help with visuals, and speed up many rounds of revision and cleanup.
What that changed
Instead of stopping at a basic workable spec, the assignment could become a richer teaching artifact with optional background pages, polished concept links, and more thoughtful scaffolding around the core code task.
More concretely, AI was useful for work such as:
- producing multiple candidate phrasings for hard ideas
- helping turn dense technical ideas into first-year-friendly explanations
- suggesting diagrams and visual structures for pages
- speeding up page editing, consistency checks, and refinement across the site
- making it feasible to keep improving the assignment after the minimum viable version already existed
At the same time, important things were deliberately not handed off:
- deciding what students should and should not implement
- deciding what is pedagogically appropriate for a first-year algorithms subject
- deciding what counts as good code structure and good reasoning
- deciding when an explanation is actually clear rather than merely fluent
- deciding what tradeoffs matter between realism, complexity, workload, and fairness
Why students benefit
The result is not just “an assignment with AI somewhere in the workflow”. The point is that the assignment can be better in ways students actually feel.
It can be more interesting because it ties a real modern idea to a manageable programming task. It can be more realistic because collaborating with AI is now part of real technical work. It can be more pedagogically valuable because the surrounding support material can be deeper, more varied, and more responsive to likely confusion points.
It also creates room for something that is often hard to achieve at once:
- a task that still expects genuine student reasoning
- a spec with stronger conceptual support
- a presentation that feels more like a serious technical artifact than a rushed worksheet
That combination is the real reason for the collaboration. The goal was not novelty for its own sake. The goal was to make a better assignment.
One Additional Note
One additional note: Kacie thinks the skills loss that can come with AI use is real, and she should not be blamed for any AI-related mishaps attached to this assignment. Those can be blamed on Dr. Cohney.