Week 4: The Toolbelt: LLMs, Agents and Other Ways to Not Be the Author

Week 4 — a survey of everything that lets you start an assignment the night before it is due and still hand in something with a straight face

15 March 2027

Hicks, Humphries and Slater argue that a language model’s output fits Frankfurt’s definition exactly: produced with no regard for truth, only for plausibility. This week takes that as an operating manual rather than a warning, and tours the stack — LLMs, agentic scaffolding, templates you can gut and refill — noting where each tool’s seams show.

Outline#

• a working definition of “last minute” as a design constraint, not a personal failing
• what each tool is actually good at, versus what it’s good at sounding good at
• where the seams show, and why that’s the marker’s job to find, not yours to hide
• next week: putting the whole stack in front of a real deadline