AI has positioned itself as a force multiplier in Instructional Design.
Yet, AI in Instructional Design still has limitations.

High demands for these Designers still creates demand for even better portfolios of higher quality. Which over the past few years, has negatively impacted the trajectory for Instruction Designers due to task automation.

• Documents become presentations.
• Manuals become training content.
• Designers started generating instead of building content.

This model can only work if the cost of being wrong is negligible.

Instructional design has shifted from construction to generation.

But how viable are the outputs for areas where the cost for being wrong is expensive?

AI fails where the cost of being wrong is high.

Low Stakes vs High Stakes

Correctness matters more than speed. 

If the corporation only needs simple presentations of their documents, this is the simple task that can be automated, especially if quality is not a main reason for the content.

The environment for the product must be considered.

Consider the difference between the following 2 examples:

• Example 1:
  A corporation needing instruction for their HR course represents a low stake environment.
  
• Example 2:
  An aerospace enterprise needing training or other instructional content, represents a high stake environment.

These two categories have different structural needs and demand different regulations and margins of error. 

When generative text goes awry in the HR course, it’s a minor error. But in that high stakes environment, it could represent death or destruction of assets leading to legal liability. Errors in real world scenarios cascade to failure.

This creates a dependency on SME validation.

As is the nature of generative text, the model does very well with predicting accurate sounding advice. But just because the text sounds correct doesn’t mean it’s based in real world logic and risks.

This implies that subject awareness and expertise, particularly in more technical and high risk areas, will be sought after.

High stakes environments offer a strategic value proposition in Instructional Design.

Inputs vs Outputs

Digital Transformation and standardization create value.

AI assumes clean inputs and predictable structure. 

Most real-world documentation has neither.

Most assume existing assets can be used in the generative parsing of text. Sometimes this is true, but when changing to a different format or medium, this becomes exponentially more difficult.

The assets must follow strict rules of format, and the output must also have strict rules. 

The most compelling part is when working with multiple organizations that don’t have format standardization, especially when the industry doesn’t have a standardization. It makes systemic approaches more difficult to implement.

In essence, creating a service that assists in standardized outputs despite chaotic documentation is an essential service to perform today.

Especially if that output is using a different visual format for content. This can be videography or digital animation. This ability to transform traditional documentation into another medium is the strongest amplifier.

In practice, this means AI outputs can’t be treated as final artifacts—they require validation against system behavior, SME input, and real-world constraints before they become usable.

The best type of output is one that is consistent. And the only way to establish consistent results is to have clearly defined and categorized inputs and a systematic approach for redistributing those portions into the final product.

AI is only one component in a larger system of transformation. Other steps may include a combination of Machine Learning and image/object recognition, and/or other tools.

Communication Dependencies

Trust and feedback loops verify content.

Content outputs that require clear communication and processes are a strong differentiator against traditional documentation.

When transferring the knowledge from experienced veterans to the upcoming generation entering the field, this is a key opportunity to take advantage of. Particularly when considering the upcoming generation are digitally comfortable and experienced veterans aren’t.

The key is to become the translator between the two parties, establishing a flow of information in that system. 

That translation layer: human, contextual, validated; is what makes the system reliable.

—

The ultimate direction for AI implementation for businesses is “to create ever-improving customer experiences and drive-lower unit cost” (Rewired, Lamarre et al., 2023).

This new tool doesn’t fail everywhere, yet expertise remains indispensable. Especially when creating processes for large form decision making in large companies.

AI works best where error is tolerable.

It fails where correctness is non-negotiable.