AIforBC practical guide
AI training for BC workplaces
Build shared boundaries, role-specific skill, and 30 days of measured practice so AI use becomes safer and more useful after the workshop ends.
Direct answer
The best workplace AI training changes how approved work is performed—not only what employees know about AI.
Start with approved tools and clear information rules. Teach one repeatable method. Practise on real, role-specific tasks with safe data. Require source checking and human review. Then measure adoption, quality, time, and exceptions for 30 days.
Employees may already be ahead of formal adoption
Statistics Canada reported that, from September 2024 to July 2025, 22% of Canadian workers aged 15 to 69 had used generative AI in their main job or business during the previous 12 months. The study notes that worker-reported use was higher than the share of businesses reporting AI use in the comparable business survey, which may indicate employee-led use alongside or outside formal organizational strategies.
That gap is a reason to train and govern use now. It is not a reason to assume every employee or workplace uses AI, and it is not a BC-specific estimate.
Choose the training outcome before the format
| Need | Best starting format | Deliverable |
|---|---|---|
| Leadership alignment | Executive briefing and decision workshop | Priorities, risk posture, ownership, and next-step decision. |
| Shared responsible-use foundations | All-staff workshop plus practical reference guide | Approved tools, permitted-data rules, review expectations, and escalation. |
| Role-specific capability | Hands-on cohort organized by workflow or function | Tested examples, reusable instructions, and supervised practice. |
| Opportunity discovery | Workflow assessment with representative employees | Prioritized use cases, baselines, owners, and pilot plans. |
| Sustained adoption | Workshop plus 30- or 90-day practice sprint | Coaching, measurements, corrections, and expansion decisions. |
Seven subjects every workplace program needs
- Capabilities and limits: what current tools can do, how they fail, and why confident language is not proof.
- Approved tools and accounts: which services, configurations, and integrations employees may use.
- Information boundaries: what is public, internal, confidential, personal, regulated, or prohibited.
- A shared task method: task, context, constraints, quality test, and accountable review.
- Source discipline: finding primary evidence, distinguishing fact from inference, and recording material sources.
- Role-specific practice: exercises that resemble the actual documents, decisions, and exceptions employees face.
- Escalation and incidents: what to do when a tool produces harmful content, exposes information, fails, or creates uncertainty.
Use a three-zone information rule
| Zone | Typical information | Default training rule |
|---|---|---|
| Green | Public, non-sensitive, approved example, or synthetic information. | May be used in approved tools for suitable low-consequence tasks. |
| Yellow | Internal business information, unpublished work, or data requiring context-specific authorization. | Use only in approved configurations and workflows with clear purpose and controls. |
| Red | Credentials, highly confidential information, personal or regulated information without an approved basis, or material subject to strict client restrictions. | Do not enter into an AI service unless the organization has explicitly approved the use and controls. |
The categories must be adapted to the organization. The purpose is to give employees a usable default and an escalation path, not to replace privacy, security, legal, or professional analysis.
Practise on work, not isolated prompts
A role-specific exercise should include the original task, available evidence, allowed information, output requirements, failure cases, and review standard. Learners should compare the assisted process with the current process and record material corrections.
- Customer-facing teams: prepare a response from approved knowledge, identify uncertainty, and escalate an exception.
- Managers: prepare a decision brief that separates evidence, assumptions, alternatives, and unanswered questions.
- Operations: organize recurring reports or requests while preserving the authoritative record and approvals.
- Professional services: summarize and compare source material without inventing facts or replacing accountable professional judgment.
- Construction and field teams: structure approved notes or search controlled documents, then verify the final record.
A 30-day workplace adoption cycle
- Prepare: choose two or three roles, approved tools, low- or moderate-risk workflows, baselines, owners, and measures.
- Learn: teach shared boundaries, the task method, source checking, privacy, security, and review.
- Apply: complete supervised role-specific work and record time, quality, corrections, and friction.
- Standardize: turn useful attempts into reviewed templates or work instructions. Document exceptions.
- Decide: expand, revise, stop, or choose a different tool or workflow based on evidence.
Measure four different things
| Layer | Question | Example measure |
|---|---|---|
| Capability | Can people use the approved method? | Successful completion of a realistic task with correct source and review behaviour. |
| Adoption | Are approved workflows being used? | Users repeating the workflow correctly during the 30-day period. |
| Outcome | Did the work improve? | Full-cycle time, throughput, quality, response time, or customer outcome. |
| Risk | Did use stay inside the boundaries? | Material corrections, unapproved data, missed review, incidents, or escalations. |
If drafting becomes faster but checking and correcting consume the savings, the workflow has not yet improved.
Questions to ask a training provider
- Will the examples be adapted to our roles and workflows?
- How do you address personal, confidential, and client information?
- Do you teach source verification, uncertainty, and human review?
- What reusable materials or work instructions will remain?
- How will you accommodate different experience levels?
- What happens after the workshop?
- How will capability, adoption, outcomes, and risk be measured?
- What claims can you support with evidence, and what is still a hypothesis?
The purchasing rule
Choose a provider or program when the training reflects your work, teaches clear boundaries, produces reusable operating material, includes supervised practice, and defines how results will be measured. Avoid programs built mainly around tool hype, long prompt lists, unsupported productivity claims, or demonstrations employees cannot safely repeat.
Frequently asked questions
Answers before you take the next step
What should workplace AI training include?
Useful training should cover approved tools, permitted information, a repeatable task method, source and fact checking, privacy and security, human-review rules, role-specific practice, escalation, and measurement after training.
How long does AI training take?
A briefing can establish awareness, but durable workplace capability usually needs practice over several weeks. A practical model combines a focused workshop with 30 days of role-specific application, office hours, review, and measurement.
Should every employee receive the same AI training?
Everyone needs shared foundations and boundaries, but exercises should reflect role, access, information sensitivity, and consequence. Managers, customer-facing staff, analysts, and field teams do different work and need different practice.
How should an employer measure AI training?
Measure capability, approved adoption, workflow quality, full-cycle time including review, and incidents or policy exceptions. Attendance and satisfaction are useful signals but do not prove that work improved.
Source trail
Primary sources used in this guide
- Workplace artificial intelligence use: A profile of sociodemographic and job characteristicsStatistics Canada
Official worker-level research on generative AI use at work, including the gap between worker-reported use and business-reported adoption.
- Guide on the use of generative artificial intelligenceGovernment of Canada
Detailed public-sector guidance on accountable, secure, transparent use, training, oversight, change management, and realistic expectations.
- Principles for responsible, trustworthy and privacy-protective generative AI technologiesOffice of the Privacy Commissioner of Canada
Joint Canadian privacy-authority principles on necessity, proportionality, consent, safeguards, transparency, accuracy, access, and accountability.
- Cyber security for users of Generative Artificial IntelligenceCanadian Centre for Cyber Security
Official learning resource covering safe workplace use, limitations, ethical concerns, and cyber-security considerations.
AIforBC uses official and primary sources where practical. This guide provides general operational education, not legal, privacy, cybersecurity, or financial advice.
Need help choosing?