The Four-Frame AI Communication Template.

A generative scaffold for writing about any AI topic with accountability, equity, and the public interest in view.

Most AI discourse oscillates between hype and fear. The Four-Frame Template moves it toward accountability — for any topic, any audience, on demand. One scaffold, ten posts.

Adapted from the FrameWorks Institute's Framing the Social Implications of AI, this template is the scaffold behind the Columbia IKNS5303 (Digital Organizations) and Chautauqua "AI in Society" curricula. It produces a piece of content — essay, carousel, or keynote segment — that:

Builds public understanding instead of mystifying AI
Makes human responsibility visible instead of treating AI as an autonomous force
Surfaces who is harmed instead of speaking only in abstractions
Anchors conversation in shared values instead of technical specs

The four frames

Frame 1

Explanation — distinguish AI from human intelligence

Goal: demystify. Make the system legible without reducing it to magic or menace.

What is the AI in this story actually doing, in plain language, that a human would otherwise do — or wouldn't do at all?

Avoid: "The AI decided…" (implies agency). "The algorithm thinks…" (implies cognition). "AI just does math" (reduces too far).

Better: "It was trained on X, so it surfaces Y." "It's pattern-matching against past examples, not reasoning from principles." "A human would weigh A, B, C; this system weights what it was optimized to weight."

Frame 2

Metaphor — make human involvement visible

Goal: show that humans built it, humans deploy it, humans benefit from or are harmed by it.

What metaphor reveals the humans behind the system — designers, data labelers, deployers, decision-makers?

Useful families: Apprentice / understudy. Recipe / instrument. Lens / mirror. Subcontractor.

Retire: "Black box" (mystifies). "Brain" / "mind" (anthropomorphizes). "Robot overlord" (collapses nuance).

Frame 3

Issue — explain how and where bias plays a role

Goal: make bias concrete and traceable, not a vague accusation.

Where in the lifecycle of this system does bias enter — data, design, deployment, or downstream use?

Data bias — what was over- or under-represented in training?
Design bias — what was optimized for, and at whose expense?
Deployment bias — where is it used, who decides, who reviews?
Downstream bias — who acts on the output, with what incentives?

Frame 4

Values — show how AI affects communities

Goal: tie the conversation to shared values and the communities most affected.

Whose community bears the cost when this goes wrong, and what value is being violated — dignity, autonomy, voice, equity, safety?

Values to anchor in: Equity. Voice / agency. Dignity. Public interest. Accountability.

Avoid: abstract "fairness" without naming who is unfair to whom; individualized blame ("users should be careful") instead of structural framing.

The fill-in-the-blank scaffold

For any AI topic, spend ten minutes filling this in. Then write the closer — one sentence that ties it together and moves the reader toward accountability, equity, and the public interest.

Frame	Prompt
1. Explanation	What is the system actually doing, in plain language? How is it different from a human doing this work?
2. Metaphor	What metaphor makes the humans behind it visible? (Apprentice / instrument / lens / subcontractor.)
3. Issue	Where does bias enter — data, design, deployment, downstream? Name the specific mechanism.
4. Values	Whose community is affected, and what value (equity, voice, dignity, accountability, public interest) is at stake?

Worked example — AI hiring tools

Explanation	Hiring AI ranks résumés by matching them to patterns from past hires. It isn't evaluating fit; it's pattern-matching to whoever the company has hired before.
Metaphor	An apprentice trained by the HR team's last decade of decisions — including the ones they'd want to take back.
Issue	Bias enters at the data layer: if past hiring favored certain résumés, the system encodes that as the target to match. The recruiter often never sees the candidates it filters out.
Values	Job seekers from underrepresented backgrounds lose voice and recourse — rejected before a human sees them. The value at stake is equitable access to economic opportunity.

"The question isn't 'is the algorithm fair?' — it's 'who is accountable for what this system was taught to reproduce?'"

Worked example — AI in classrooms

Explanation	Classroom AI generates responses based on training data; it doesn't know the student in front of it. It produces plausible-sounding answers, which is not the same as correct or pedagogically appropriate ones.
Metaphor	A confident substitute teacher who's read a lot but never met the class.
Issue	Design bias: most tools are optimized for engagement and ease, not for learning or equity. Deployment bias: under-resourced schools adopt AI as a substitute for staffing, not a supplement.
Values	Students in under-resourced communities risk getting more AI and less human attention — widening, not narrowing, the opportunity gap.

"The promise of AI in education is real, but the burden of getting it right falls hardest on the communities with the least margin for error."

Use this in your work. Teach it in yours. The framework is meant to travel.

For the printable companion + the next iteration as it evolves, subscribe to Signals from the Curve.

How to cite this framework:

Reitz, C. H. (2026). The Four-Frame AI Communication Template. chrishuberreitz.com/frameworks/four-frame-template. Adapted from FrameWorks Institute, Framing the Social Implications of AI.