The landscape of learning design and development is experiencing a seismic shift. As artificial intelligence becomes increasingly sophisticated, we’re also witnessing its impact on LD space. But it is not a replacement for human expertise; rather, it is an evolution in how we work and where we use automation. AI in learning design and development is a reality and not going anywhere.
After months of trial and error, refinement, and practical application across multiple projects, I’ve put together a comprehensive framework that reimagines the traditional design models (like Addie) for the age of AI. This approach aligns remarkably well with some recent academic research, particularly the ADGIE model (Analysis-Design-Generation-Individualisation-Evaluation).
My Core Principle for Using AI in Learning Design: AI for Generation, Humans for Vision and Validation
The core principle behind this framework is simple yet transformative: AI excels at generating content, while humans excel at validating, refining, and providing strategic direction. The goal is not to replace instructional designers or subject matter experts, but to elevate their role—from content creators to pedagogical architects and quality gatekeepers.
Each project I’ve managed or delivered has its nuances: some begin with only an idea, others with a fully developed curriculum. Milestones and tasks vary depending on scope, while the chosen combination of tools and platforms influences the workflow. Yet, at a high level, this is the consistent approach I follow when using AI to power learning content generation and authoring.
AI in Learning Design and Development: Six-Phase Framework
Let me walk you through the complete process. It consists of six major phases with built-in validation checkpoints throughout.
Phase 1: AI-Powered Source Content Generation
This phase begins with preparation. The Project Manager or Coordinator creates and distributes instructions – let’s call it a source content pack. This comprehensive brief should include the module descriptor, program overview, target learner personas, and crucially, AI-enhanced prompts designed to help SMEs or ID structure their thinking.
The SME or ID then drafts high-level source content, including:
- Module content synopsis
- Case studies and practical examples
- Use cases and project scenarios
- Recommended resources
Then, using AI prompts, generate a comprehensive, structured version of the SME source content.
Validation Gate #1 follows – this is a comprehensive review, checking for:
- Content completeness and accuracy
- Alignment with learning outcomes and awarding body standards
- Quality of examples and case studies
It can be done by someone in the team responsible for pedagogical alignment – e.g. Learning Manager (LM), Lead Learning Designer.
This gate is critical—if content doesn’t pass, we return to the SME before any AI storyboarding or authoring begins. This prevents downstream amplification of errors.
| Summary: |
|
Phase 2: AI-Driven Storyboard Development
This is where AI truly begins to shine. Using the validated SME content, the ID leverages AI prompts to:
- Analyse and structure the module flow
- Generate a draft storyboard using optimised AI prompts
- Plan assessments aligned with pedagogical requirements and standards
- Integrate case studies throughout the narrative
- Identify interactive element requirements for external tools
- Add resource requirements and references
Notice the human role here: the ID is prompting, curating, and structuring, not writing from scratch. The AI generates; the human orchestrates and validates.
Validation Gate #2 – Lead Learning Designer ensures:
- Pedagogical soundness of the structure
- Assessment alignment with learning outcomes
- Format compatibility with the target LMS
- Technical feasibility of proposed interactive elements
| Summary |
|
Phase 3: Detailed Content Expansion with AI
Primary Responsibility: Instructional Designer (ID)
Accountability: Lead Learning Designer
Using the approved draft storyboard, AI prompts generate a fully detailed, expanded narrative including:
- Complete storyboard with learning sequences
- Detailed MCQs and assessment items with rubrics
- Elaborated case studies with context
- Detailed simulations and branching scenarios
- Assignment briefs and project specifications
- Assets and images requirements
What’s crucial here is the iterative refinement. The ID doesn’t accept first drafts generated by AI—they validate, refine prompts, and regenerate until the output meets quality standards.
Validation Gate #3 is perhaps the most comprehensive, checking:
- Pedagogical quality and instructional design principles
- Assessment alignment and quality
- LMS compatibility
- Resource integration feasibility
Phase 4: Content Generation and Authoring with AI
Here we have the Branching Point—Platform-Based AI-powered generation vs. Traditional LMS capabilities.
This is where our framework acknowledges a critical reality: not all learning platforms are created equal, and your development path varies significantly based on your technology.
Scenario 1 – LMS-Native AI Content and Structure Generation + External Content Development
4A LMS-Native Content Generation
For most platforms, even those that claim their AI generates all content, content generation and assembly happen sequentially:
- Generate module structure and core content using AI prompts and files (storyboard content files) optimised for the specific LMS.
- Configure learning sequences and prerequisites
- Generate or upload and configure assessments.
- Add case studies and interactive activities.
- Upload resources and external references.
Validation Gate #4a – review content quality and formatting in LMS, navigation and learning flow
4B External Content Development
Running in parallel with 4a, this phase handles content that requires specialised external tools to be created:
- Interactive activities and simulations
- External interactive elements like Articulate
- Video production with AI tools like Synthesia
For all of them, we need to identify requirements, finalise scripts, produce and edit, and ensure accessibility (captions, transcripts)
Validation Gate #4b ensures quality, pedagogical value, and accessibility compliance and integration testing in the LMS environment
Scenario 2 –AI Content Generation and Authoring within the LMS
- Generate module structure and core content using AI prompts and files (storyboard content files) optimised for the specific LMS.
- Configure learning sequences and prerequisites
- Generate and configure assessments.
- Generate case studies and interactive activities.
- Interactive activities and simulations
- Generate AI Avatar videos with tools like Synthesia.
Validation Gate #4 – review content quality and formatting in LMX, navigation and learning flow, pedagogical value, and accessibility compliance
Some real-life examples
- Traditional LMS (Canvas LMS): Manually build the LMS structure, author all content (images, videos, interactivities) elsewhere and upload it, then assemble.
- AI-powered LMS (Cypher, Rise Up, Thrive): Generate structure and flow using AI; generate basic elements (standard blocks—text, sometimes images; some platforms allow generating AI videos); author multimedia and interactivities elsewhere and upload them; finalise by assembling the course.
- AI-superpowered LMS with native content generation and advanced content authoring: I came across several platforms that claim to do it all, but I was really impressed with only one.
- Sana Labs AI is by far my favourite. Sana’s content generation and authoring capabilities enable Phases 4a and 4b to be collapsed into a single streamlined process. Content generation, interactive activity authoring, and assembly happen within the platform itself, dramatically reducing complexity and timeline. The platform’s AI handles much of the formatting and structure, allowing designers to focus even more heavily on pedagogical quality and learner experience.
- Coursebox AI is another platform that does more and lets you generate the structure and some interactive content right within it. Still, from the UX and learner experience point of view, it is way behind Sana.
Phase 5: Course Assembly and Set Up
Primary Responsibility: Instructional Designer (ID), Project Manager (PM)
Accountability: Lead Learning Designer
The reality is that even with the most sophisticated and superpowered platforms, there are going to be elements that need to be brought into the platform and assembled – images, infographics, icons, to name a few. This phase brings everything together:
- Add all externally developed learning objects
- Configure gamification and engagement features
- Set up learner support elements
- Configure tracking, analytics, and reporting
- Create Alpha version documentation
- Set up test learner accounts
Validation Gate #5 is a comprehensive checkpoint before stakeholder review.
Phase 6: Review, QA, Go Live
Primary Responsibility: Lead Learning Designer, QA Team
Accountability: Project Manager (PM)
This final phase includes:
- Learning Design Review by the LLD and SME
- Functional QA testing by the QA team
- Consolidation of all findings and recommendations
- Creation of Module QA Record for accreditation
- Alpha Approval Gate—the final decision point
What else is important in AI enhanced learning design and development?
AI in Learning Design Redefined Roles (New RACI)
In my AI in Learning Design framework, I use a clear RACI model (Responsible, Accountable, Consulted, Informed). Here is an example:
The nature and remit of these roles have fundamentally shifted and evolved in the AI-powered environment.
Traditional role – content reviewer and pedagogical expert
AI-enhanced role: Chief validator, pedagogical architect, and quality gatekeeper
- Accountable for: All validation gates and final approval decisions
- Focus: Ensuring AI outputs meet pedagogical standards, align with learning science, and serve learner needs
Traditionally, ID is a content creator and storyboard writer
AI-enhanced role: Prompt engineer, content curator, and AI orchestrator
- Responsible for: Generating content through AI, refining prompts, and validating outputs
- Focus: Crafting effective prompts, curating source materials, and iterative refinement
Traditionally, a content provider and reviewer
AI-enhanced role: Strategic content architect and accuracy validator
- Focus: Providing high-quality source content and validating technical accuracy
- Enhanced by: AI-guided prompts that help structure their expertise
Traditional role: timeline, resource and budget management.
AI-enhanced role: Orchestrator of human-AI workflows, ethical oversight
- Focus: Managing the complex interplay between human validation and AI generation
Traditional role: Testing functionality and finding bugs
AI-enhanced role: Testing human-AI handoffs, ensuring accessibility, validating AI-generated assessments
- Focus: Functional testing plus pedagogical validation of AI outputs
AI powered LD Practice is Validated by Research
What’s remarkable is how closely this framework—developed through practical trial and error—aligns with other academically proposed models, e.g., the ADGIE model. Both recognise that:
- AI excels at generation, not strategy: The ADGIE researchers found that while 94% of professionals want AI to generate materials, only 49% felt ready to let AI select core teaching methods. My validation gates embody this caution.
- Continuous validation is essential: ADGIE makes evaluation “transversal and continuous.” My five validation gates throughout the process operationalise this principle.
- Human expertise elevates: As Dr Philippa Hardman notes in her analysis of ADGIE, the role shifts “from content creator to curator & validator” and “from project manager to pedagogical expert.” My RACI model reflects exactly this transformation.
- The human remains in the driver’s seat: The overwhelming insistence on human validation of AI content isn’t just a preference—it’s a pedagogical necessity.
The Human Factor is More Important Than Ever
Here’s the counterintuitive truth: introducing AI into learning design doesn’t diminish the human role—it elevates it.
In traditional workflows, instructional designers spend enormous time on:
- Formatting and restructuring content
- Writing first drafts of assessments
- Creating basic variations of similar content
- Manual copy-editing and consistency checks
With AI handling these tasks, humans can focus on what we do best:
Strategic Thinking
| Pedagogical Expertise
|
Quality Assurance
| Ethical Oversight
|
The validation gates throughout our framework aren’t bureaucratic checkpoints—they’re intentional moments for human expertise to shape, refine, and ultimately approve what AI generates.
Lessons Learnt about AI in Learning Design
Through implementing this framework across multiple projects, several key insights have emerged:
1. Prompt engineering is now a core competency. Instructional designers, project managers, LDs, and SMEs need to develop sophisticated prompting skills. The quality of AI output is directly proportional to the quality of the prompt.
2. Source content quality matters more than ever. Garbage in, garbage out. The SME content phase is critical because AI amplifies what it receives—both quality and flaws.
3. Validation gates prevent expensive rework. Early validation saves massive time downstream. Catching issues before content is generated into multiple formats is exponentially more efficient.
4. Platform choice dramatically affects workflow. AI-native platforms like Sana Labs collapse development phases that require weeks in traditional LMS environments.
5. The human-AI handoff is the critical junction. Most failures occur not in AI generation or human validation alone, but in the handoff between them. Clear processes and documentation are essential.
What’s in the Future of AI in Learning Design
This framework represents a snapshot of my practice in late 2024 and 2025. As AI capabilities evolve, so too will my processes. But some principles will remain constant:
- Human judgment on pedagogical strategy
- AI acceleration of content generation
- Rigorous validation throughout the process
- Ethical oversight of learner data and experience
- Continuous iteration and improvement
The goal isn’t to build learning experiences faster for the sake of speed—it’s to create better learning experiences by allowing human experts to focus on what matters most: understanding learners, applying learning science, and ensuring every design decision serves learning outcomes.
As I continue to refine this approach, I don’t think the question whether AI will transform learning design—it already has. The question is whether we’ll thoughtfully integrate it in ways that enhance rather than diminish the quality of learning experiences we create.
- Hilali, A., & Chergui, F. (2025). “Toward a New Instructional Design Methodology in the Era of Generative AI”—A foundational academic paper introducing the ADGIE model and discussing AI’s evolving role in learning design.
- Dr Philippa Hardman’s “Beyond the Hype: What 18 Recent Research Papers Say about How AI Is Transforming Learning Design”—A comprehensive synthesis of contemporary studies on AI in instructional design.
- Whatfix’s “ADDIE Model: 5 Stages of Instructional Design (2025)”—An updated overview of the classic ADDIE model for comparison and context.
- EdTech Books “A Co-Creative Approach for AI-Enhanced Instructional Design”—Insights and practical models for co-creation between human experts and AI in learning development.
- Digital Learning Institute “Education Technology Trends to Watch in 2025: 10 Innovations“—A look at cutting-edge tools, platforms, and pedagogies shaping contemporary learning environments.
- Learning Science Digest “The Learning Research Digest vol. 34”—Curated updates and new studies relevant to AI and learning design.
Note, that this post provides general information about AI in Learning Design.
It is important always to consider the specific context and requirements of your learning projects. If you have any questions or would like to delve deeper into the topic, please email me or book a free online consultation via my contact page.
Make sure to check out my other posts related to planning online courses, designing and developing learning content and delivering training. I share strategies and tools that you can use and many practical tips.
