AI is quickly becoming part of the instructional design workflow. But while the conversation around AI in Learning & Development often focuses on speed, many teams are still figuring out what “good” actually looks like in practice.
Because faster content creation alone does not guarantee better learning.
In many organizations, instructional designers are already under pressure to produce more content, support more stakeholders, localize for more audiences, and prove business impact with fewer resources. In our 2026 AI in Frontline Enablement report, training creators said the most time-consuming part of their workflow is building lessons themselves - structuring content, drafting copy, and turning raw information into something usable.
At the same time, frontline employees are frustrated by repetitive, generic training experiences that feel disconnected from the reality of their jobs. 48% say training “repeats stuff I already know,” while 59% want learning tailored to their role, level, or specific knowledge gaps.
That tension explains why AI is gaining traction in instructional design.
The opportunity is not simply to create more content faster, but to reduce low-value production work so instructional designers can spend more time creating learning experiences that are contextual, practice-driven, and tied to real capability development.
Most training creators are balancing far more than course authoring alone.
They are expected to gather knowledge from SMEs, structure learning pathways, write copy, source visuals, localize content, distribute lessons, evaluate performance, and report on impact. Many of those workflows are still heavily manual.
The result is predictable: there is less time available for the parts of instructional design that actually improve learning quality.
Our research found that training creators often skip embedding job-based scenarios, adding visual polish, creating custom media, and revisiting existing lessons because they are under pressure to keep producing new content.
This is where AI is becoming useful.
For many instructional designers, AI is helping reduce friction in areas that traditionally slow content production down. Teams are using it to turn SME notes into draft lessons, simplify long-form content, adapt training for different audiences, generate assessments, localize material, and speed up first-pass creation work.
72% of training creators said they would like AI to transform SME call notes directly into draft lessons.
That matters because SME collaboration is currently one of the biggest bottlenecks in training creation. 79% of training creators say sourcing information from SMEs slows content production down significantly.
The strongest instructional designers are not using AI to replace design judgment. They are using it to spend less time wrestling with production workflows and more time improving relevance, practice, reinforcement, and application.
Many AI-generated learning experiences fail for the same reason many traditional learning experiences fail: they prioritize content production over capability building.
The pressure to move quickly is understandable. Training creators are already juggling SME interviews, localization, reporting, stakeholder approvals, media editing, platform formatting, and endless review cycles. Large parts of that process are still manual.
That workload is shaping how AI gets used in practice.
Spend five minutes in an instructional design forum or Reddit thread and the same patterns show up repeatedly: designers pasting source material into ChatGPT, asking for draft learning objectives or quiz questions, then manually rebuilding everything inside tools like Articulate Rise or Storyline afterward. The AI speeds up the drafting process, but the operational friction surrounding the workflow still remains.
And because many prompts are still fairly broad, the outputs often end up broad too.
A request like:
“Create a 10-minute lesson on workplace safety”
does not tell the model:
So the result is often informational content that reads cleanly, but feels detached from the realities of the job.
That challenge came up repeatedly throughout a recent session hosted by eduMe Learning Designer Sasha Howard, in a Workday-delivered session on how to build capability at scale.
One of the strongest themes was how easily learning pathways drift toward content progression rather than capability progression. Employees complete modules. Knowledge gets distributed. Yet organizations are often left with very little visibility into whether someone can consistently apply that knowledge in practice.
AI can accelerate production dramatically. But instructional designers still need to shape the conditions for practice, reinforcement, application, and demonstration.
If you want to explore the full session, you can watch Building Capability at Scale: A New Model for Skills Development here.
The instructional designers getting the best results from AI are usually approaching it less like a content vending machine and more like a collaborative design assistant.
Instead of asking AI to simply “build training,” they are using it to support different stages of the learning process. That might mean generating role-specific scenarios, adapting content for different audiences, simplifying operational information, creating reinforcement activities, drafting manager coaching prompts, or building examples tied to real workplace situations.
The throughline across all of these use cases is contextual relevance.
That matters because frontline employees are increasingly expecting training to adapt to them personally. The same research cited earlier found that 93% of frontline workers want training that “learns from them” and adjusts over time.
How attainable or not this exact state is today is less relevant - the deeper takeaway is that employees want guidance more reflective of their role, their existing knowledge, the situations they actually encounter, and the pace at which they work.
AI gives instructional designers far more flexibility to create those kinds of experiences at scale, though it must be paired with a stronger capability-building framework underneath.
You've likely heard it a million times by now: "garbage in, garbage out". In other words - the quality of AI-generated output depends heavily on the quality of the prompt itself.
Strong prompts provide instructional context, audience detail, learning objectives, and performance expectations. Weak prompts produce generic filler.
Below are several prompt approaches instructional designers can use to improve output, and by extension - learning quality.
These prompts help tailor learning to a specific audience or role.
Example prompt:
“You are designing onboarding training for newly hired warehouse associates working overnight shifts in a high-volume distribution center. Rewrite this content using clear, operational language appropriate for employees with less than 30 days tenure.”
This tends to produce more realistic, relevant outputs than generic rewriting prompts.
Scenario generation is one of the strongest use cases for AI in instructional design.
Example prompt:
“Create three realistic customer service scenarios where a retail associate must de-escalate a frustrated customer while following company return policy.”
This helps instructional designers move beyond informational learning into practice-based learning.
AI can also help simplify overly dense material.
Example prompt:
Reduce cognitive overload in this SOP by simplifying language, removing unnecessary detail, and restructuring information into a step-by-step workflow suitable for mobile learning.
This is particularly useful when adapting technical or operational documentation.
AI can support stronger capability assessment workflows.
Example prompt:
Generate five observational checklist criteria a frontline supervisor could use to verify whether an employee can correctly perform this machine safety procedure during a live shift.
This moves learning closer to measurable performance.
AI is useful for generating post-training reinforcement content.
Example prompt:
Create three short reinforcement questions that test whether a hospitality employee can apply this service recovery process during a busy check-in period.
These prompts support retrieval practice and long-term retention.
Many organizations need training adapted for different audiences and markets.
Example prompt:
Adapt this lesson for frontline manufacturing employees in Mexico. Preserve technical accuracy while simplifying language and improving cultural relevance.
This can dramatically reduce localization effort.
💡 If you want additional examples, exercises, and templates, our free AI Prompt Workbook for Learning & Development includes practical prompt structures instructional designers can adapt for real workplace training workflows:
AI can accelerate content creation, it can support brainstorming, and it can reduce administrative burden.
But it still lacks operational context.
An AI model does not understand:
This is why human review remains critical.
In our research, 67% of training creators said AI support cannot come at the cost of accuracy, context, or tone.
Instructional designers still need to validate outputs, refine tone, add operational nuance, ensure realism, and align learning to business priorities.
The strongest AI-assisted training workflows still rely heavily on human instructional judgment - AI does not supplant expertise.
Many organizations already produce large amounts of training content.
The challenge becoming harder to ignore is whether employees can apply it consistently in real work. This becomes especially relevant in frontline environments, where one employee's ill-compliance (e.g. failure to operate a machine safely) can have a dramatic ripple effect.
As a result, frontline organizations increasingly need learning experiences that reinforce behavior, improve operational consistency, support performance under pressure, help employees self-serve information quickly, and build measurable capability over time.
This is one reason many instructional designers are shifting attention away from completion metrics alone. Completion simply relays an employee was exposed to training material - it does not relay whether or not they can action a process smoothly and safely.
Ensuring capability exists is especially pertinent in operational environments where employees need to apply procedures accurately, handle unpredictable situations, follow safety standards consistently, and make decisions under time pressure.
AI has the potential to support stronger capability-building here - particularly when paired with role-specific guidance, reinforcement, adaptive learning, and real-world practice scenarios.
AI has fundamentally shaken up how L&D teams can approach daily design workflows.
But where AI is best used here is not as a blind creation machine, but as a tool that removes enough production friction so that instructional designers can spend more time on things like: improving relevance, practice, reinforcement, and capability development.
This shift in approach to AI usage matters especially in frontline environments, where employees need support that is contextual, accessible, and directly connected to the realities of the job.
If you want to experiment with AI-assisted training creation yourself, eduMe’s AI Course Creator allows teams to turn existing documents, SOPs, and training materials into mobile-ready microlearning in minutes.