Using Images with LLMs for Better Instructional Design | Alphabees

Education leaders face a dual challenge: they must continuously evolve their methods while integrating new technologies into existing workflows. Large Language Models offer capabilities that extend far beyond pure text processing. The ability of modern AI systems to interpret visual inputs opens new perspectives for instructional design – from analyzing handwritten notes to evaluating complex work environments.

For L&D teams and educational institutions in the DACH region, this represents an opportunity to make the design process more efficient without replacing human expertise. The visual analysis capability of LLMs can complement and accelerate existing workflows – provided it is deployed purposefully and responsibly.

Visual Artifacts as a Starting Point for Learning Design

In the daily work of instructional designers, numerous visual documents are created: whiteboards from workshops, storyboard sketches, photos of work environments, or screenshots of existing learning platforms. These artifacts contain valuable information that previously often had to be manually transferred and structured.

With image-processing LLMs, these visual inputs can be directly analyzed. A photo of a whiteboard from an SME meeting can be examined in seconds for core topics, potential learning objectives, and action steps. The AI identifies connections and suggests structured derivations – a process that would take considerably more time manually.

Handwritten notes from planning phases can also serve as input. The AI extracts concepts, organizes thoughts into logical structures, and generates initial module drafts. The key principle: outputs should always be understood as a starting point to be validated and refined by subject matter experts.

Analyzing and Optimizing Existing Courses

Screenshots of learning platforms and course modules offer another practical application. Education leaders can photograph or capture existing e-learning content and have an LLM review it for optimization potential.

The analysis can encompass various dimensions:

Cognitive Load:

Is the information density appropriate, or does the layout overwhelm learners?

Accessibility:

Are contrasts, font sizes, and navigation structures designed to be accessible?

Interaction Design:

Does the module offer sufficient opportunities for active engagement with the learning material?

User Guidance:

Is the navigation intuitive and does it support the learning process?

This analysis provides concrete starting points for improvements that can be implemented in the next iteration. Particularly for universities and continuing education providers managing large course portfolios, this approach can significantly accelerate quality assurance.

Translating Real Work Environments into Learning Scenarios

A particularly valuable application lies in developing scenario-based training. Photos of real work environments – whether a hospital room, a production hall, or an office workspace – can serve as the foundation for authentic learning situations.

When an LLM analyzes such images, it can derive realistic decision scenarios, safety situations, or typical challenges that employees experience in their daily work. These context-based scenarios increase the relevance of training and promote transfer of learning to practice.

Physical teaching materials can also be digitized this way. Anatomy models, technical diagrams, or laboratory setups can be photographed and analyzed. The AI then suggests ways these hands-on learning experiences can be translated into digital simulations or interactive modules.

Using Data Visualizations for Strategic Decisions

L&D teams regularly work with dashboards, evaluation reports, and survey results. Screenshots of these data visualizations can also serve as LLM input. The AI helps identify patterns in the data and derive possible recommendations for action.

For decision-makers, this means faster interpretation of learning data and more efficient derivation of measures. Which modules show high dropout rates? Where do knowledge gaps appear? What interventions could improve learning outcomes? Visual data analysis through AI can address these questions and support strategic decisions.

Responsible Use of Multimodal AI

Despite all efficiency gains, critical review of AI outputs remains essential. LLMs can make errors, misinterpret connections, or deliver incomplete analyses. Human expertise – the specialized knowledge of instructional designers, understanding of the target audience, comprehension of organizational contexts – remains central to high-quality learning offerings.

Data protection aspects are equally important. Photos of real work environments or people should only be used with appropriate authorization. Educational institutions should establish clear guidelines for handling visual data in AI systems.

For universities, academies, and companies in the DACH region, integrating multimodal AI into the design process offers significant advantages. AI tutors that are directly embedded in existing learning management systems like Moodle can leverage these analytical capabilities to improve both course design and individual learning support.

The combination of visual analysis and intelligent learning support enables more efficient and effective design of the entire learning cycle – from conception through delivery to evaluation. What remains crucial is that technology is understood as a tool that supports human expertise, not replaces it.

Frequently Asked Questions