A chief learning officer guide to AI-enabled knowledge work

Classify the CLO's role

The CLO may own the learning strategy and programme design, influence workforce priorities, review evidence quality, and collaborate with functional managers who own the work. Budget, technology, security, and employment decisions may sit elsewhere. A portfolio should name the CLO as owner, influencer, reviewer, or collaborator for each decision instead of assuming one executive controls the full system.

Start by naming the population, work context, business question, and accountable operating owner. “Improve AI skills” is too broad. “Help proposal teams identify unsupported claims in an AI-assisted compliance review” is narrow enough to design learning and evidence around.

Build a progression tied to work

The UNESCO AI Competency Framework for Teachers organizes capabilities across five dimensions and three progression levels: acquire, deepen, and create (UNESCO teacher framework). A workplace programme can borrow the idea of explicit dimensions and progression without claiming that the education framework validates a corporate curriculum.

For each objective, define an observable behavior, practice activity, evidence artifact, reviewer, and boundary. Distinguish knowledge of a rule from applying it under realistic constraints. Distinguish applying a technique once from adapting it safely to a new context.

Sources: [1]

Separate the evidence layers

Maintain four layers. Participation includes attendance and completion. Learning includes knowledge or skill demonstrated in a bounded exercise. Transfer includes behavior observed in a new but comparable task. Organisational outcome includes measures such as quality, review burden, cycle time, or incident rate. Each layer needs its own evidence and interpretation.

Do not infer transfer from satisfaction or an organisational outcome from a single exercise. Record missing evidence, selection effects, task differences, and changes in tools or policy. If managers contribute observations, define what they saw and when rather than collecting only a confidence rating.

Worked artifact: a verification learning portfolio

Imagine a synthetic four-week programme for proposal professionals. The portfolio maps three behaviors: locating governing requirements, tracing claims to sources, and escalating contradictions. Participants first annotate a short synthetic request for proposal. They then practise on a different evidence pack with an AI drafting tool. The final task asks them to produce a correction memo under time pressure.

The portfolio records the source packs, task difficulty notes, allowed assistance, observable checks, reviewer rubric, and missing-evidence rule. A participant who produces no contradictory claim is not automatically credited with verification; the task must create a fair opportunity to encounter and handle one. Managers receive follow-up questions, not a pass threshold or employment recommendation.

Verify the programme logic

CDC's 2024 programme-evaluation framework emphasizes context, evidence gathering, conclusions, and acting on findings as connected parts of evaluation practice (CDC programme evaluation framework). Use a simple chain: need, activity, expected behavior, evidence opportunity, interpretation, and decision. Test every link for an unsupported leap.

Before scaling, review task relevance with functional experts, accessibility with representative participants, evidence consistency with calibrated reviewers, and data handling with the appropriate owners. Revisit the portfolio when the role, tool, policy, evidence pack, or decision changes.

Sources: [2]

Preserve the boundary

A learning evidence portfolio can support a more disciplined conversation about programme design and observed work. It does not establish causal impact, mature benchmarks, performance outside the task, or control effectiveness. Two Prune can create a controlled opportunity to observe AI-enabled work and produce task-bounded evidence for human interpretation. The customer retains responsibility for learning, workforce, and employment decisions.