In 2023, AI was touted as the answer to nearly every business problem on the planet — including the pesky job of writing documentation.

Two years down the road, the AI hype frenzy has peaked, and people are taking a more realistic look at its limits and the ongoing need for human involvement in content creation.

In this post, we’ll examine the pros and cons of excluding subject matter experts (SMEs) from your company’s documentation process.

The Downside of Involving SMEs in Technical Writing

Let’s start by discussing the reasons company management might feel tempted to dispense with SMEs altogether when generating technical documentation.

• Availability — A technical writer working on a critical document is often at the mercy of a busy SME’s schedule. Predicting when someone can free up enough time for an interview varies widely, and it can’t be reduced to an exact science.
• Time — If a technical writer sends a document to an SME for written review, the turnaround time may be extremely slow. Once again, the SME’s other duties typically take precedence, potentially bringing the writing project to a grinding halt.

The Upside of AI-Generated Documentation

Many business managers initially hoped to use AI to increase speed and cut costs associated with creating documentation. By removing the human element, AI’s efficiency becomes obvious and offers some clear advantages:

• Turnaround Time — AI can generate copy at a dizzying rate compared to what a human writer can produce. It speeds up the writing process exponentially.
• Research — Whether collecting new research or organizing scattered notes, AI’s efficiency shines.
• Initial Drafts — While a human writer might struggle with a topic and require several passes to get a feel for the material, AI can pare down nonessential information and instantly create a tight first draft.
• Brainstorming — Humans are known to suffer from writer’s block occasionally. AI can help by generating brainstorming suggestions to stimulate new ideas.

The Limits of AI-Generated Documentation

Despite its many advantages, AI falls short when it comes to documentation precision. While it can execute some time-driven tasks quite well, relying on AI exclusively to generate accurate content carries several risks:

• Outdated Information Models — Much of the data used to train today’s AI systems reflects information current only through 2020, making some of the material used to generate 2025 text already obsolete.
• Product Knowledge and AI — Although AI is fairly competent when tackling general research topics, its ability to delve into highly specific and detailed subject matter remains limited.
• Best Industry Practices — Just as the general data used to train AI may be outdated, the best industry practices embedded in its training materials might no longer apply.
• Hallucinations — When AI encounters a topic it doesn’t fully understand, it sometimes produces irrational or incorrect answers known as hallucinations. These responses may be logically consistent within AI’s narrow framework of training data but fail real-world scrutiny.
• Predictions — AI struggles to predict likely outcomes because it can’t anticipate situations with no past precedent.
• Storytelling — Some documentation requires examples to illustrate a point, but AI has no personal memory of specific situations that would help readers grasp the material more clearly.

Management Misperceptions

When AI was first introduced for content creation, many managers had only a vague understanding of its limitations.

“My biggest issue with AI right now is that management (at my company, at least) seems to think that it’s magic and that it can increase my output enough to make up for us being short-staffed.”

“The technology isn’t yet able to do what they want it to, but they don’t listen. I’m not sure if it’s because of the hype around AI or just wishful thinking.”

“The CTO scheduled calls with me about automating all technical content with AI. I was open to the idea, but the technology was not good enough at the time… There is a quality to human-generated content that cannot be replicated by machines.”

“There are some uses for AI-generated content. But it’s not good enough for what my company wants it to do.”

“A regulated industry, such as energy or biotech, will make using AI impossible because of compliance regulations and the amount of contextual knowledge you need retained within groups that are siloed.”

General Versus Specific Knowledge

The final comment above highlights a major issue AI cannot solve. Beyond the confidential information held within regulated industries, many organizations depend on proprietary processes to run their business.

These processes are unique to individual organizations and cannot be extracted from the general knowledge base available to AI.

To develop documentation for these specialized processes, technical writers must first map those processes to determine the appropriate structure. Mapping a custom process requires logic, critical thinking, and SME interviews to fill in the gaps.

Senior managers impressed with AI’s speed often fail to recognize the essential partnership between the technical writer and the SME in producing contextually accurate documentation.

What SMEs Bring to the Mix

Because AI cannot yet be trusted to perform the heavy lifting of content creation, companies must continue to rely on human oversight in AI-generated documentation.

Whether leaders fully appreciate it or not, both SMEs and technical writers are necessary to validate critical documentation. These experts bring value far beyond what AI can offer:

• Contextual Depth — SMEs not only understand their topic but also how it relates to other critical steps in a process. Their holistic understanding surpasses what AI can provide.
• Hands-On Experience — SMEs bring years of practical experience, often more current than the information models used to train AI.
• Ability to Judge Content Validity — AI can present facts, but those facts may reflect only a narrow slice of the subject. SMEs and technical writers can identify gaps or inconsistencies.
• Capacity to Forecast Trends — AI cannot offer projections without historical precedent. SMEs can anticipate technological, regulatory, or market changes that may shape future needs.
• A Human Voice — SMEs can share anecdotes and examples that humanize content and help readers relate to it.
• Ability to Build Trust — The personal insight an SME provides lends credibility and resonance to the material.
• Human Ingenuity — Beyond fact-checking, SMEs generate new ideas and solutions that AI cannot. Their understanding extends beyond existing data to original thought.

Will SMEs Soon Become Obsolete?

While AI currently lacks the creative problem-solving and contextual understanding needed for high-quality documentation, future advances may narrow that gap. It’s reasonable to ask whether SME interviews may eventually become unnecessary.

Current data suggests that SMEs are safe for the foreseeable future — and possibly far beyond. The Future of Jobs Report 2025 lists the top ten most sought-after skills in the global workforce for the next decade. Notably, half of these skills cannot be automated:

From this list, it’s clear that the human element will remain a foundational part of tomorrow’s workforce.

When it comes to creating written content, AI may bring speed and efficiency, but it must be guided by human expertise. SMEs will continue to play a central role in producing clear, accurate, and reliable documentation for a long time to come.