Introduction and acknowledgment

Robin welcomed participants to the Sprout Labs webinar focused on using AI to speed up instructional design. He introduced himself as the founder of Sprout Labs and host of the Learning While Working podcast.

He acknowledged the Traditional Owners of the land he was joining from—the Muwinina people. He noted with sadness that the group did not survive colonisation and paid respects to their elders past, present and emerging.

Robin also briefly explained his movement disorder, generalised dystonia, to explain why his head may move involuntarily during the session. He linked this to neuroplasticity and his personal experience with adapting to change.

Sprout Labs overview

Robin described Sprout Labs’ core work as helping L&D professionals create effective learning through platforms and capability development. He mentioned:

• The Glasshouse learning experience platform
• Being a Totara partner
• Resources like the podcast, blog, webinars and ebooks

The webinar served as a preview of the AI for smarter learning design lab, launching in May.

Session framing

Robin outlined three core areas covered in the session:

• Using AI to support and speed up design tasks
• Developing effective prompting techniques
• Creating learner profiles and focusing on real learner needs

He emphasised that these insights were drawn from podcast interviews, Sprout Labs’ own use of AI, and early prototypes with clients.

Participant reflections and aims

Participants were invited to share in chat what they hoped to gain. Responses included:

• Improving content creation and workflows
• Understanding prompt design
• Exploring fast instructional design
• Staying current with AI developments
• Guidelines for AI-supported design without full automation

Robin used these inputs to adapt the session to the group.

Whiteboard activity: AI in current workflows

Participants marked their current engagement with AI. Responses ranged from enthusiastic use to limited adoption due to workplace restrictions or lack of skills.

Robin noted that Sprout Labs had also restricted use of generative AI on codebases due to data security concerns.

How large language models work

Robin explained how models like ChatGPT operate:

• Words are mapped as numbers based on patterns and probability
• These models predict likely word sequences based on past data
• Transformers are the core architecture used in most models
• Reasoning models (e.g. Gemini) take a step-by-step approach, good for maths and logic but poor at writing
• Mini models are smaller and better suited to low-powered systems

He highlighted issues such as:

• Bias
• Repetition of common knowledge
• Fabricated references
• Superficial learning design recommendations

Working alongside AI

Robin introduced the metaphor of AI as an intern—helpful but needing supervision. He discussed using AI for:

• Summarising meetings
• Generating transcripts
• Text-to-speech tools to speed up writing (especially helpful due to his dyslexia and dystonia)
• Helping users focus on high-value tasks rather than routine content generation

Identifying superpowers and weaknesses

Participants used annotation tools to identify their strengths (superpowers) and weaknesses (kryptonites) across different media types:

• Written language
• Visual communication
• Voice and music
• Moving images
• Verbal language

Robin encouraged participants to lean into their strengths and use AI to support weaker areas. He shared his own experience of dyslexia and reduced motor control, noting how AI tools like text-to-speech have improved his workflow.

He gave examples of tools that can support media production:

• Image generation: MidJourney, DALL-E, Canva
• Voice and audio: ElevenLabs, Descript
• Video: Runway, HeyGen, Sora
• Transcription: Otter.ai, MS Teams, Krisp

Using AI to support team capabilities

Robin shared a client example where an L&D team focused on improving their analysis phase with AI. They developed prompts and systems to generate deeper, more thoughtful learning solutions. This shift helped them save time in other parts of the process and make better design decisions.

He emphasised that using AI to fill gaps—individually or at team level—can lead to faster and more effective learning design.

Prompt engineering basics

Robin provided a “crash course” in prompt design, with key principles:

• Be specific: Clearly define what you want the AI to produce
• Provide context: The more background information you give, the better the results
• Coach and refine: AI rarely gets it right the first time; treat it like a draft and adjust as needed
• Use structure: Frameworks like “act as...” help guide responses

He shared an example of using ChatGPT to:

• Create a scenario about screen sharing risks in virtual meetings
• Generate a 150-word story
• Build a multiple choice question with feedback for each option

Robin noted that the results were a good draft, though further editing was often needed to make the learning more realistic and aligned.

Prompting as a dialogue

Robin and participants reflected on how natural (or not) formal prompts feel. Some participants said they start with a structured prompt and shift into a conversational tone as they refine. Others said they use techniques like:

• Asking ChatGPT to wait before responding
• Requesting it ask clarifying questions first
• Feeding it back previous responses to improve consistency

Colloquialisms and ambiguous terms were identified as common causes of confusion. Participants agreed that refining prompts was often a helpful way to clarify their own thinking.

Designing with learner personas

Robin highlighted that many learning designs miss the learner’s voice. He shared a story where a subject matter expert claimed to also speak for learners, which didn’t sit right with him.

To improve this, he demonstrated how to use AI to build learner personas, starting with:

• A prompt to list seven key features of a good persona
• A follow-up to write short personas based on those features
• An example of “Priya,” a nurse in the Emergency Department, based on #learner1

Robin encouraged participants to use these AI-generated personas throughout the design process, such as:

• Reviewing storyboards
• Giving feedback on modules from a learner’s perspective

He showed how reworking scenarios to fit specific personas (like Priya) can increase relevance and engagement.

Recommended resources

Robin shared two curated mailing lists that he finds valuable for staying current with AI developments:

• Superhuman by Zain Kahn: A daily newsletter with summaries of key AI news.
• Dr Philippa Hardman’s Substack: A weekly newsletter focused on the intersection of learning and AI.

Robin highlighted a post from Dr Hardman where she critiqued AI-generated SCORM content and called for stronger learning design practices.

Promotion of the AI for smarter learning design lab

Robin wrapped up by inviting participants to take the next step through the upcoming AI for smarter learning design lab. Key features include:

• Starts 1st May
• Five 2-hour live Zoom sessions
• Curated AI prompt library
• Monthly AI community of practice
• Designed for instructional design workflows
• Real-world projects and hands-on practice
• Discounts for Sprout Labs clients and teams

Session wrap-up and participant reflections

To close the session, Robin asked participants to share how they plan to apply what they learned. Responses included:

• Creating WHS scenarios
• Generating learner personas
• Refining prompts for reuse
• Building content prototypes
• Designing learning pathways
• Improving analysis using AI, not just content generation

Robin acknowledged these reflections and praised the group’s openness in sharing throughout the session.