And what eLearning designers need to do next
Generative AI is no longer an experimental tool. It’s embedded in our daily work. Employees are using AI to draft emails, summarize reports, create documents, explain policies, prepare presentations, and respond to customer inquiries. But what does this change actually mean for e-learning professionals?
An extensive study by Microsoft Research provides some useful clarity. In Working with AI: Measuring the Applicability of Generative AI to jobs (Tomlinson, Jaffe, Wang, Counts, & Suri, 2025), researchers analyzed 200,000 anonymized conversations with Microsoft Copilot and mapped them to real-world work activities using the O*NET framework. Rather than predicting future disruption, this study looked at how AI is already being successfully used in workplace operations. The findings reveal important implications for the use of generative AI in workplace learning that instructional designers, L&D managers, and digital learning teams should take note of.
1. AI is most effective for information-based jobs
The study found that AI performs best in activities related to:
Writing and editing content. Provides step-by-step instructions and technical details. Teach or clarify a concept. Gathering and organizing information. Communication with customers and stakeholders. Prepare instructional materials and information materials.
In other words, AI is good at information tasks such as creating, processing, and communicating information.
Here’s why this is important for e-learning: Almost every job includes information tasks. Even in operational or front-line roles, documentation, reporting, communication, scheduling, or compliance accounting is required. The scope of AI is not limited to technical roles. It cuts across industries. This means AI competency development should not be siloed in IT training. It should become part of your core learning strategy.
2. The real skill shift is cognitive, not technical.
One of the most useful distinctions in research divides the impact of AI into two types:
AI that assists employees (augmented) AI that performs parts of the task itself (delegation)
Some roles use AI as a productivity partner. Some delegate certain components of their work to AI systems. For e-learning professionals, this distinction changes the way courses are designed. Most current AI training focuses on:
Tool walkthrough. A quick tip. Function description.
But research shows that’s not enough. What employees really need is support such as:
Decide when to use AI. Evaluate the AI output. Detection of incomplete or inaccurate responses. Risk and escalation management.
In other words, in workplace learning, it is necessary to not only train how to use generative AI but also to train judgment skills.
3. Completion rates do not prove AI readiness
The researchers measured the impact on AI based on:
Task completed successfully. The scope of AI capabilities in work activities. Applicable to the real world beyond the profession.
It was not measured how many people “completed the training.” For eLearning teams, this is a wake-up call. If your AI initiative success metrics include:
Course completion rate. satisfaction score. Login frequency.
You may be measuring engagement rather than impact. Some of the more meaningful metrics include:
The quality of decision making has improved. Reduce rework. Enables faster response while maintaining accuracy. Improved escalation decisions. Improved documentation clarity.
AI will change the way we work. Learning metrics should reflect changes in task performance.
4. Why basic knowledge still matters
This research suggests that AI could help democratize access to expertise. When used effectively, AI can empower employees to perform tasks previously assigned to experts. However, this benefit is only realized if users are able to critically evaluate the AI’s output. Without basic knowledge, employees may do things like:
Accept inaccurate answers. Miss the contextual nuances. Unable to detect hallucinations. Apply the guidance incorrectly.
This creates a new instructional design priority that blends AI skills with enhanced domain knowledge. AI competency training should include:
Validation framework. Error detection checklist. Promote risk awareness. Reflective decision-making questions.
The goal is confidence in the calibration, not blind trust.
5. Where AI is struggling right now (and why it matters)
The study also found that AI is less effective in the following areas:
Physical or manual work. Highly situational or complex decision making. Specific analysis tasks.
This reinforces an important design principle: AI should be positioned as a support tool rather than a replacement for professional judgment. Training should enable learners to understand:
Limits of AI. Situations that require human supervision. If escalation is necessary. How to combine AI output with contextual insights.
This prevents over-reliance and builds responsible usage habits.
Practical implications for eLearning professionals
So how should learning teams respond? Here are five shifts you can make.
1. Design role-specific AI learning paths
Avoid general AI awareness courses. Instead:
Identify high-frequency information tasks by role. Map where AI meaningfully overlaps. Build targeted learning modules for those moments.
for example:
Sales team → AI-assisted proposal creation + verification Human resources team → AI-assisted policy communication + compliance check Operations → AI-assisted documentation + report clarification
The relevance of generative AI use cases in workplace learning has led to its increasing adoption.
2. Use scenario-based e-learning instead of passive modules
AI capabilities cannot be learned through slides alone. Integrate:
Scenario branching. Decision-based simulation. Risk assessment exercise. Output evaluation activity.
Have learners review AI-generated content and decide:
Is this accurate? What am I missing? What risks does this pose? Will it escalate?
This builds applied abilities.
3. Incorporate AI into performance support, not just courses.
The AI itself works as follows.
On-demand explainer. I’m a writing assistant. Feedback partner. Summary tool.
Integrate AI into your workflow rather than isolating it into training sessions. example:
Provides a prompt library within the LMS platform. We provide a practice environment that utilizes AI. Generate adaptive feedback using AI.
We support learning in the flow of work.
4. Competency Framework Updates
Traditional competency models rarely include:
AI collaboration skills. Rapid improvement capabilities. Validating the output. Risk adjustment.
These must be incorporated into modern digital literacy frameworks. AI fluency is becoming part of professional competency.
5. Redefine the role of the instructional designer
Here’s an unpleasant reality. AI can already draft course outlines, write objectives, generate quiz questions, and summarize small business interviews. If instructional design remains focused solely on content creation, its value diminishes. Opportunities exist in the following places:
Performance diagnosis. Workflow adjustments. Simulation design. Behavioral measurements. Human-AI interaction design.
Moving from content creation to performance engineering increases the strategic value of L&D.
final thoughts
A study by Microsoft Research does not predict that AI will eliminate jobs. Instead, it shows where AI overlaps with real-world work activities today. This overlap is significant and increasing.
For e-learning professionals, the question is no longer whether to teach AI skills. The real question is: Are we designing learning to improve human judgment in AI-enhanced tasks?
Because successful organizations are not the ones that deploy the most AI tools. They will be the ones training their employees to use AI thoughtfully, critically and strategically. And that starts with how we design learning now.
