Skip to content

AI + Learning

AI is a partner, not a replacement for judgment.

After nearly two decades in learning design, AI feels less like another authoring tool and more like a new layer of capability. I use it to accelerate research, simplify complex information, prototype faster, generate media, test ideas, and explore learning experiences that were previously too expensive or time-consuming to build.

The important part is knowing where AI helps — and where human judgment still has to lead.

01 — AI-Assisted Learning Design

Where the model learns its place in the process.

I treat AI as a collaborator inside an established instructional design process, not as an automatic course generator. The goal is speed without giving up accuracy, context, learner empathy, or professional judgment.

01

Research Acceleration

AI can rapidly synthesize source material, stakeholder input, documentation, transcripts, and existing training so I can identify patterns, gaps, questions, and learning opportunities faster.

Human validation still determines what becomes part of the solution.

02

Content Analysis

I use AI to examine source material for duplication, missing information, reading complexity, structure, terminology, and opportunities to simplify content before development begins.

Useful for finding signal in large amounts of information.

03

Objective Development

AI helps generate, challenge, and refine learning objectives, scenarios, practice opportunities, and assessment ideas before I apply instructional judgment and finalize them.

The model generates options. The designer decides what matters.

04

Scenario Generation

Generative AI makes it possible to explore many realistic scenarios, decision paths, learner personas, and edge cases quickly.

Especially useful during early design and SME review.

05

Rapid Storyboarding

I use AI to move quickly from rough concepts to outlines, scripts, visual sequences, interface ideas, and prototype-ready content.

This reduces production friction without skipping design thinking.

06

Assessment Development

AI can accelerate first-pass question writing, distractor generation, scenario variations, feedback language, and assessment mapping.

Every final assessment still requires human review for accuracy, fairness, and instructional value.

02 — Generative Media

Media that used to require a crew can now begin with an idea.

One of the most exciting changes in learning design is the collapsing distance between an idea and finished media.

Generative video, voice, imagery, animation, and rapid editing tools now allow small learning teams to create experiences that would previously have required larger budgets, longer schedules, or outside production resources.

My multimedia background makes this especially powerful because I can evaluate AI-generated media using the same standards I would apply to traditional production: composition, pacing, editing, sound, visual continuity, instructional purpose, and audience fit.

Cinematic Generative Video

Exploring tools such as Veo, Seedance, QuickFrame, Pictory, and other emerging platforms to create realistic scenes, visual explanations, mood pieces, demonstrations, and concept-driven learning media.

AI creates possibilities quickly; traditional editing still shapes the final experience.

AI Voice & Audio

Using tools such as ElevenLabs, Synthesia, WellSaid, and related platforms for narration prototyping, voice exploration, localization concepts, and scalable audio production.

Human review remains essential for tone, pronunciation, context, and appropriateness.

AI Imagery & Visual Design

Generating concepts, backgrounds, illustrations, textures, interface references, visual metaphors, and presentation assets that can then be refined through traditional design workflows.

Useful for rapid visual exploration and prototyping.

Rapid Multimedia Prototyping

Combining generative video, voice, imagery, screen capture, traditional editing, motion graphics, and instructional design to prototype complete learning experiences much earlier in the design cycle.

The goal is not “AI content.” The goal is better learning, produced faster.

03 — Multimodal Learning

One subject. More than one way to understand it.

AI makes it increasingly practical to transform the same source material into multiple learning formats without rebuilding everything from scratch. A single body of validated content can become:

01Written explanation
02Visual summary
03Diagram
04Scenario
05Interactive practice
06Audio explanation
07Short-form video
08Long-form walkthrough
09Job aid
10Knowledge check

The opportunity is not simply to create more content. It is to give learners different ways to enter the same idea based on the task, context, accessibility need, or preferred learning experience.

04 — Agents + Knowledge Systems

The next learning interface may not look like a course.

I am increasingly interested in learning systems that can retrieve, reason over, and respond to organizational knowledge instead of forcing employees to search through static course catalogs and document libraries. This includes experimentation and working knowledge around:

AI agentsCustom assistantsMCP-enabled workflowsRetrieval-based knowledge systemsPrivate document repositoriesGemini / NotebookLM-style knowledge environmentsCustom GPT-style assistantsStructured organizational knowledge basesSearch-driven learning experiences

AI Agents

Designing and experimenting with task-focused agents that can support activities such as research, content analysis, knowledge retrieval, workflow assistance, coaching, and learning support.

The value comes from clear scope, trusted data, and good guardrails.

MCP-Enabled Workflows

Exploring Model Context Protocol concepts and workflows that allow AI systems to securely interact with approved tools, repositories, and services rather than relying only on information inside a conversation.

This creates interesting possibilities for enterprise learning and knowledge work.

Private Knowledge Environments

Exploring tools and architectures where AI works against approved organizational sources such as internal documents, repositories, learning libraries, and curated knowledge bases.

  • NotebookLM-style environments
  • Private document libraries
  • Custom repositories
  • Retrieval-based assistants
  • Company-specific learning copilots

The model should work from trusted organizational knowledge whenever possible.

Learning Copilots

Prototyping concepts where employees can ask questions, locate resources, receive explanations, compare information, and discover relevant learning directly through conversational interfaces.

This is part of the thinking behind concepts such as Company Brain.

05 — Enterprise Responsibility

Powerful does not mean careless.

The more capable AI becomes, the more important it is to understand what information should — and should not — be shared with a model.

For enterprise learning, privacy and security are not side issues. Training teams routinely work with proprietary processes, internal documentation, employee information, customer data, regulated content, and intellectual property.

My approach is to design AI-assisted workflows around approved tools, trusted repositories, appropriate access controls, and clear human review.

01

Data Boundaries

Do not place confidential, regulated, proprietary, or personally identifiable information into public AI systems unless the organization has explicitly approved that use.

02

Approved Platforms

Use enterprise-approved AI environments, private repositories, and organizational accounts whenever sensitive internal knowledge is involved.

03

Trusted Sources

Ground AI-assisted learning in validated company documentation, SME-approved content, controlled repositories, and known sources rather than allowing the model to invent authoritative information.

04

Human Review

AI output should be treated as a draft, recommendation, or synthesized interpretation until a qualified human validates it.

05

Least Necessary Access

Agents and connected AI workflows should only receive access to the tools, repositories, and information required for the specific task.

06

Traceability

When possible, learning systems should allow users and reviewers to understand where information came from and what source material informed the response.

06 — AI Workflow Experiments

Tested, documented, and honest about limits.

Prototype

Course Build Accelerator

A prototype workflow using AI to help move from SME and source material into first-pass course structure, learning objectives, draft activities, assessments, scripts, and media concepts.

Designed to accelerate the blank-page stage, not automate instructional design.

ChatGPTClaudeInstructional Design
Workflow

Scenario Stress Test

A workflow for generating alternate learner scenarios, edge cases, likely mistakes, and decision paths before final scenario-based learning is developed.

Useful for pressure-testing a design before development begins.

AIScenario DesignHuman Review
In practice

Narrated Walkthrough Production

An AI-assisted workflow supporting the production of narrated portfolio and training walkthroughs, including script development, narration options, captions, editing support, and media preparation.

Combines AI acceleration with traditional video production skills.

Generative MediaVideo ProductionEditing

07 — AI Product Lab

Two learning concepts, plus side experiments.

These are prototypes, MVPs, and product concepts built to explore how AI, search, UX, automation, and knowledge systems could change the way people learn and find information.

They are experiments, not production enterprise deployments.

Additional product experiments

These projects are outside traditional L&D, but they allow me to experiment with AI-assisted development, UX, databases, search, automation, mapping, payments, and product thinking.

The technology changed. The responsibility didn't.

AI can help me research faster, prototype sooner, generate media that once required a production crew, build new kinds of learning interfaces, and explore ideas that would have been impractical only a few years ago.

But the job is still to understand what someone needs to know or do — and create the clearest, most useful way to help them get there.

That part remains human.