PAN Lab example

The same AI under full guardrails: the professional office

Same AI, full guardrails: verification time is budgeted, writes are gated, records carry provenance labels, and someone with authority reviews the deployment on a schedule. In the sociotechnical simulation — a modeled office, not a real one — errors stick about 16% of the time. Two tests even strong offices face open this round: staff quietly pasting case details into unvetted consumer AI, and a vendor model update that shifted behavior under controls tuned to the old version — and because the model runs on the vendor's host, that same update is a second way client data can slip past the boundary. The lesson is diminishing returns — and what it costs to keep controls from drifting.

Comparative teaching networkCaseworker documentation & copilots

Open this example in PAN Lab v0.1 to apply pressures and levers and watch what the system does.

What this models

This example runs on the High-governance professional office network: 5 components and 12 pathways between them. Every context in the Lab is a stylized model, never a reconstruction of any actual deployment, and each assumption behind it carries a provenance label.

Evidence base: 3 assumed · 1 published baseline. In the Lab, the shaded evidence band behind each headline readout draws its width from the least-established class below.

baseline
How often errors stick in this office (≈16%) comes from the sociotechnical simulation, not from measuring a real office.
assumed
The strong peer-consultation pathway reflects the professional-standards norm of seeking a colleague's advice on difficult calls; its regime relief is an authored assumption in the Lab's qualitative vocabulary.
assumed
The model's raw error output is held identical across all three office cultures; only oversight differs.
assumed
Professional-standards controls are assumed to be followed as designed, with no drift at baseline.

Sources and evidence

What this example rests on, claim by claim. Every entry resolves to the same ledger the Evidence Registry publishes.

In the sociotechnical simulation, the same AI in three modeled office cultures - stylized, not real workplaces - let errors stick at very different rates: roughly 75% under low-oversight autonomy, 20% under human supervision, and 16% under high-governance professional controls.
scenarioillustrative PAN-run result
No published source is attached to this claim yet.

Where this connects

Institutional pressures in this domain

Workload surge — Demand outruns staffing; per-case attention shrinks and review becomes triage.
Deadline pressure — Statutory or managerial timeliness rules reward fast approval of machine output over slow disagreement.
Reviewer bottleneck — One fixed-capacity checking stage sits between AI output and consequence; everything queues behind it.
Staff turnover — Experienced skepticism leaves; new staff calibrate their trust on the tool itself.
Vendor opacity — The deploying institution cannot inspect the model, data, or update pipeline it is accountable for.
Compliance over substance — Paper controls (sign-offs, checklists) satisfy audits while the behavior they describe erodes.

All of them in context on the Caseworker documentation & copilots domain page.

Levers available here and the patterns behind them

Upgrade model — Improve the model
Review on schedule — Oversight cadence & retrospectives
Keep skills sharp — Deskilling-arrest mandate
Understand the system — Understand the system
Require sign-off — Conformity assessment gate
Gate vendor updates — Vendor quality gate
Keep prompts neutral — Framing and mirroring reduction
Vet connections — Connection authorization
Store less data — Data minimization
Assign a challenger — Structured dissent