SMASH 2026 · Paper #22 · Result card
Damned if You Do(n't): How Epistemic Technology Spoils Expertise
The default radiologist case reads as shunned expertise. Two judges coded 36 comments with κ = 0.93.
case radiologistprofession Radiologist at a community hospitalstatements 4corpus 36 comments (default)judge A claude-haiku-4.5judge B claude-sonnet-4.6
Your case
Shunned expertise
Radiologist at a community hospital; adoption, others · second judge: Slopped (disagree) · 5/5 quotes verified verbatim
Damned if you don't
Outdated
the quadrant the same audience points to had you made the opposite adoption choice
Two-judge agreement (κ)
0.93
95% CI [0.82, 1.00] on 36 comments · stance κ 0.89, locus κ 0.96
Most common form (both judges agree)
Shunned
8 of 36 comments · default corpus
From the abstract
We advance a concept of spoiled expertise which takes distinct forms based on two dimensions: knowledge workers' stance of adoption of epistemic technology, and locus of evaluation of expertise. Together these two dimensions culminate into four forms of spoiled expertise: (1) slopped expertise, (2) lagging expertise, (3) shunned expertise, and (4) outdated expertise.
Evidence
Paper vs. your run
| Metric | Paper | Your run | |
|---|---|---|---|
| Slopped expertise (adoption, self) | This form captures experts who adopt epistemic technology and see their own expertise devalued by such adoption. Experts internalize the discourse tha… | A 7 · B 7 · agreed 7 | corpus counts |
| Lagging expertise (non-adoption, self) | This form captures experts who do not adopt epistemic technologies, but who internalize societal discourse that valorizes the capabilities and outputs… | A 5 · B 5 · agreed 5 | corpus counts |
| Shunned expertise (adoption, others) | This form captures experts who adopt epistemic technologies, and are stigmatized by others for doing so. This adoption is perceived by the expert's au… | A 8 · B 8 · agreed 8 | corpus counts |
| Outdated expertise (non-adoption, others) | This form illustrates experts who do not adopt epistemic technologies and are stigmatized by others for this resistance. These non-adopters become sym… | A 9 · B 7 · agreed 7 | corpus counts |
| Transitions | Slopped → Shunned; Outdated → Slopped (§4) | Shunned → Slopped (extrapolation (relabelled: not a transition the paper describes)); Shunned → Outdated (extrapolation) | 'extrapolation' = not in the paper |
| Empirical measurement | none (conceptual paper; proposes an empirical research agenda) | κ = 0.93 between two LLM coders | codability, not prevalence |
| Agreement: clear-cut vs deliberately ambiguous comments | — | 97% vs 83% | 30 vs 6 comments |
| Judges vs the corpus author's intended labels | — | κ A 0.96 · B 0.96 | synthetic corpus |
Try next
- The lawyer preset — a non-adopter — The lawyer refuses AI entirely; expect Lagging or Outdated and compare how locus and mirror flip relative to the radiologist.
{"case": "lawyer"} - Same-model bias test — judge_b small — Both judges become Haiku 4.5; if κ rises sharply, part of the 0.93 reflects shared model reasoning rather than codebook clarity.
{"case": "radiologist", "judge_b": "small"} - Bring your own case — custom profession — Replace the preset with your own profession, AI use and peer reactions to see where your situation sits in the 2×2.
{"case": "custom", "profession": "Clinical pharmacist", "ai_use": "I use an AI drug-interaction checker for every complex patient and document every override.", "peer_reactions": "A physician told me the tool is making me a button-pusher, not a pharmacist || I worry I have forgotten how to reason through polypharmacy without the alert list || The pharmacy director praised my override documentation at the last audit"}
Read this before sharing
- An interpretive aid, not a measurement.
- The comment corpus is synthetic and both judges are Claude models.