What it is
A skill that a tool performs for you fades. A worker who has only ever reviewed drafts loses some of the ability to write an assessment from the start. That worker also loses some of the ability to see what a draft got wrong. Protected practice without the system keeps that ability alive, through manual casework rotations or cases worked before the tool is opened. The research cited below names trained verification skill, not general knowledge of AI, as the counter to over-reliance.
What it pushes on in the Lab
In the Lab, this lever holds down operator deference drift, the slide from checking the system's output to accepting it by default. It also raises the checking and correcting that the people using the system can do.
You can also pull this lever at a strong tier, which costs more. At the strong tier, each effect below the Lab's strongest setting pushes harder.
In the modes that offer aiming, you can aim this lever at particular parts and pathways of a network. Otherwise it applies to the whole network.
Its pattern in the Practice Library
The Practice Library describes the pattern behind this lever:Deskilling-arrest mandate
The pressures it answers
These pressures list this lever among the levers that answer them:
- Workload surges
- Staff turnover
- Autonomy expands
- AI-literacy gap widens
- Conformity spreads
- Expectations outrun the gains
A lever answers a pressure when it pushes the other way on something the pressure pushes on.
Where you can pull it
Networks in the Lab that offer this lever:69
- A contact centre's generative-AI agent assist
- A heavy-industry predictive-maintenance deployment
- Albert France Services
- Allegheny Family Screening Tool
- Allegheny Hello Baby
- Allegheny Housing Assessment
- Amazon fulfillment-centre algorithmic management
- Amsterdam Smart Check
Every network that offers it
- Arkansas ARChoices / ARIA
- Audi press-shop inspection
- Benefits Data Trust wind-down
- BMW AIQX inspection
- Caddy adviser copilot at Citizens Advice
- Calgary Drop-In Centre
- CDTFA Axyom Assist
- CHAI (chronic-homelessness prediction)
- CNAF benefit-fraud risk score (France)
- CORA, the DC CFSA policy assistant
- CrimSAFE criminal-record tenant screening
- Douglas County Decision Aide
- DWP Whitemail Insights and Vulnerability Scanner
- Eckerd Rapid Safety Feedback
- Frida (NAV Norway)
- Gaggle Safety Management
- GDS Microsoft 365 Copilot cross-government experiment
- GetCalFresh
- GitHub Copilot at ZoomInfo
- Gladsaxe model
- Homebase Risk Assessment Questionnaire
- Illinois DCFS Augintel
- Illinois Rapid Safety Feedback
- Imagine LA Benefit Navigator copilot
- Insight Bristol / Think Family Database
- IRS collection chatbots
- Kaiser Permanente Suicide-Risk Model
- LA County Homelessness Prevention Unit
- Learned Hand AI clerk pilot (LA and Riverside courts)
- Limbic Access (NHS Talking Therapies)
- London's Strategic Insights Tool
- LyssnCrisis counselor QA at ProtoCall Services (988)
- Magic Notes (Beam)
- NarxCare
- Nava assistive benefits chatbot
- Netherlands childcare-benefits scandal (Toeslagenaffaire)
- Nevada DETR generative-AI unemployment appeals
- New Zealand MSD Predictive Risk Modelling
- NJ AI Assistant
- Oregon Safety at Screening
- Oxevision camera monitoring on NHS mental health wards
- Predict-Align-Prevent
- RealPage revenue management
- Rotterdam welfare-fraud risk model
- SafeRent Tenant Screening Score
- Samagra Vedika
- San Jose's camera car
- Santa Clara County Homelessness Prevention System
- Sistema Alerta Niñez (Chile)
- SSA Insight
- The same AI under full guardrails: the professional office
- The same AI with a human checking: the supervised office
- Trelleborg's Welfare Robot
- Udbetaling Danmark data-driven control (Denmark)
- UK DWP Universal Credit Advances fraud model
- UPS delivery route optimization
- US Birth Match
- Vanderbilt VSAIL suicide-risk alert
- VI-SPDAT
- Workforce Australia Targeted Compliance Framework
- Xantura OneView (predictive homelessness flagging)
The evidence behind its effects
The Lab cites these claims from the evidence registry for this lever's effects.
A validated collaborative-AI metacognition scale (planning, monitoring, evaluation of one's own reliance) predicted collaboration benefits incrementally beyond general metacognition — verification-skill training, not generic AI knowledge, is the calibrated counter to over-reliance.[2]
Sidra, S., & Mason, C. (2026). Generative AI in Human-AI Collaboration: Validation of the Collaborative AI Literacy and Collaborative AI Metacognition Scales for Effective Use. International Journal of Human–Computer Interaction, 42(7), 5084–5108. https://doi.org/10.1080/10447318.2025.2543997
doi.org/10.1080/10447318.2025.2543997
Appears in: Evidence reverification (2026); Paramerge authored research
Topics: ai-safety, human-ai-interaction
Bucinca, Z., Malaya, M. B., & Gajos, K. Z. (2021). To Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW1), Article 188. https://doi.org/10.1145/3449287
Appears in: Evidence reverification (2026)
Topics: human-ai-interaction