Talk Is Cheap
Source: Nick Potkalitsky Substack Author: Nick Potkalitsky Original source: https://nickpotkalitsky.substack.com/p/talk-is-cheap Published: 2026-05-14 Source type: essay
Private backup: the full article text is archived in the private repository at archives/articles/nickpotkalitsky-substack-com-talk-is-cheap.source.md. It is not published on the public Quartz site.
Summary
Nick Potkalitsky argues that many school responses to generative AI in assessment are discursive rather than structural. Traffic-light systems, AI-use tiers, declaration forms, and policy warnings communicate expectations, but they usually leave the underlying assessment task unchanged and depend on voluntary student compliance. Drawing on Corbin, Dawson, and Liu, he argues that these approaches create an enforcement illusion: institutions appear to have governed AI use, while the actual conditions of assessment validity remain fragile.
The article points toward structural assessment redesign instead. Structural responses change the task, format, sequencing, or evidence environment so that desired behavior emerges from the design rather than from unenforceable instructions. Potkalitsky names process-oriented assessment, oral examinations, authenticated checkpoints, evidentiary chains across a unit, and Black Box Assessment as examples. He also connects this to disciplinary AI literacy: roles such as Critic, Verifier, Interlocutor, Editor, and Architect remain discursive if merely named, but become more structural when embedded into task sequences and classroom routines.
Pull quotes
Discursive limits
“All three are operating at the level of communication, and communication, however clear, however detailed, however earnest, cannot do what only structure can do.”
AI traffic lights
“A red-coded assessment is only red if the student decides to treat it that way.”
Enforcement illusion
“When assessment validity rests on student compliance with unenforceable rules, we are not protecting the integrity of our credentials so much as assuming it, quietly, at scale.”
From named roles to design
“Named as orientations and communicated to students, they look discursive. Embedded in how tasks are sequenced across a year, built into the conditions under which independence becomes possible rather than merely expected, they begin to function structurally.”
Durability note
The article is durable because it gives the wiki a clean vocabulary for distinguishing AI policy language from assessment redesign. Its value is not the immediate examples of traffic-light frameworks, but the larger distinction between communicating rules and creating learning environments where process, competence, and integrity are built into the structure of the task.
Big ideas
- District AI work is a long-term redesign project
- Learning still needs some struggle, even when AI can make things easier
- AI literacy should help people notice how AI changes what counts as knowing
- AI literacy has to be taught inside real subjects
Claims
- AI-assisted homework requires redesign, not just policing
- In an AI world, assessment should focus on watching students think
- Take-home essays are no longer reliable evidence by themselves
- Schools should separate AI literacy work from assessment integrity work
- AI-generated text can make finished writing less trustworthy as evidence
Key evidence and examples
- Potkalitsky distinguishes discursive AI governance from structural assessment redesign, borrowing from Corbin, Dawson, and Liu’s assessment scholarship.
- He treats traffic-light AI policies, AI Assessment Scale-style permission tiers, declaration forms, and institutional guidance as largely discursive because they communicate rules without changing the task conditions.
- He argues that these approaches can produce an enforcement illusion: the visible language of governance suggests security that the assessment design itself does not provide.
- He frames generative AI and assessment as a wicked problem, where policies and guidelines are predictable institutional responses but insufficient by themselves.
- He identifies structural alternatives including process-oriented assessment, oral examinations, authenticated checkpoints, evidentiary chains across a unit, and Black Box Assessment.
- He connects structural assessment reform to disciplinary AI literacy, arguing that AI-use orientations become more than vocabulary only when embedded into task sequencing and independence-building routines.
Education relevance
Highly relevant for AI assessment policy, academic integrity, district AI guidance, assignment redesign, disciplinary AI literacy, and professional learning. The article helps schools distinguish between rules that ask students to comply and assessment structures that make learning, process, and competence visible.