Mapping AI Adoption Stances
Given a statement, policy, or quote about AI/automation adoption, classify it along two axes:
- Who draws the line? — an individual, or a collective/community
- How absolute is the line? — total refusal, conditional/practicable exceptions, or no line at all
Then place it in one of six positions:
| Position | Who draws line | Absoluteness | Signature move |
|---|---|---|---|
| Amish | Collective | Negotiated, can flex | Community votes on a threshold |
| Old Order | Individual | Near-total | Personal renunciation ("I will not use AI") |
| Mennonite | Individual | Conditional clauses | "I'll use it for X, never for Y" |
| Luddite | Collective (movement) | Total, aimed at everyone | Organizes, unionizes, demands regulation |
| Romantic | Individual | None (but grieves) | Consents while mourning the craft |
| English | Individual | None | Full adoption, no resistance |
Progress:
- Identify the speaker/actor (individual vs. community/institution)
- Identify the mechanism (personal choice, vote, decree, strike/movement)
- Identify the scope of the line (self only, members only, everyone)
- Identify whether exceptions exist (conditional clauses = Mennonite-like)
- Check for emotional valence (grief/mourning = Romantic; defiance = Luddite)
- Assign position on the continuum
- Note exit cost — how easy is it to leave this community/stance? (low exit cost = volatile, fast-reversing positions; high exit cost = durable schisms)
- Flag uncharted-territory complications (see below) if the case doesn't cleanly fit
Uncharted territory checks
Some cases resist clean classification — flag these explicitly rather than forcing a fit:
- The Empty Chair: can the actor's stance even be practiced given entangled toolchains/infrastructure built on the thing they're refusing? (e.g., refusing AI while hosting on AI-dependent infra)
- Local inference / proximity to benefit: does the automation benefit the displaced party directly (in their own hands/home) rather than replacing them for someone else's profit? This changes the moral valence of resistance.
Example 1: Input: "Our platform votes every year on whether to allow AI-assisted submissions; this year members voted 358-144 to restrict heavy LLM use, but enforcement relies on self-disclosure, not scanning." Output: Amish. Collective actor (platform members), mechanism is a vote, line is negotiated/non-absolute (no active enforcement/scanning), scope applies only to members of this specific community. Exit cost: high (migrating a project's home), so this schism will be durable.
Example 2: Input: "I quit using AI entirely. I found I could no longer explain my own code without asking it questions first." Output: Old Order. Individual actor, near-total personal renunciation, scope is self only, no conditional exceptions, motivated by loss of craft-understanding rather than political organizing.
Example 3: Input: "I use AI to help me understand ffmpeg flags, but I never generate prose, art, or video with it." Output: Mennonite. Individual actor, conditional/practicable clause distinguishing permitted use (technical assistance in a domain they understand) from forbidden use (creative generation).
Example 4: Input: "A subreddit moderator unilaterally bans any AI-related project, with no vote, citing inability to detect AI translation; the rule reverses within days under user backlash." Output: Decree-based collective line (not a true Amish vote — more like unilateral purity-testing), but crucially: low exit cost (one click to a rival subreddit) predicts rapid reversal — contrast with Codeberg's high-exit-cost durability.
- Always separate who decides from how strict. A community can draw a loose line (Amish) or an individual can draw a total one (Old Order) — don't conflate collective with strict.
- Check exit cost whenever comparing two seemingly similar bans/schisms — it predicts durability and reversal speed far better than the stated rationale does.
- Distinguish Romantic (consents + mourns) from Luddite (refuses + organizes) — both can sound nostalgic, but only one takes collective action to change terms or pace.
- Track movement along the continuum over time for the same individual/community — positions are not static; cite direction (e.g., "moved from English toward Old Order") rather than treating classification as permanent.
- When citing this framework, prefer direct quotes that reveal mechanism ("we voted," "I will not," "I use X for Y but never Z") over paraphrase — the classification hinges on the speaker's own framing of agency and scope.
- Use the 1812 parallel (hand-loom weavers, power looms, repricing of labor-hours) sparingly as explanatory shorthand, not as proof — flag it as a falsifiable analogy, not settled law.
- Don't label a strict personal stance "Luddite" just because it's severe — Luddite requires a collective/movement aimed at everyone, with explicit asks (regulation, unionizing). A strict individual stance is Old Order.
- Don't assume community bans are durable — check exit cost first. A subreddit rule and a code-forge policy are not equivalent even if textually identical.
- Don't ignore the "Empty Chair" problem when someone claims a totally AI-free stance — check whether their toolchain/hosting/compiler dependencies quietly violate the claimed purity.
- Don't treat the continuum as one-directional — explicitly check and report whether movement is toward or away from automation; both happen.
- Don't force every case into the six buckets — infrastructure-entanglement and local-inference scenarios are explicitly uncharted; say so rather than mis-stretching a label to fit.