Instructional designers are creating content faster than ever with the help of generative artificial intelligence (AI) tools. As a result, writing is more polished, grammatically correct and consistent than ever before. Yet, many designers are noticing something missing: authenticity.
Learners are beginning to pick up on the patterns, too. The writing is clean, but it lacks a certain spark. We’ve replaced clunky writing with polished writing. But did we accidentally lose the human voice along the way? Let’s break down the hidden risks of unchecked AI writing and how to keep your eLearning content feeling human.
Risk #1: The Tone Misses the Mark
AI tools can be helpful, but sometimes can miss the mark, defaulting to language that feels unnatural. The tone can swing from overly cheerful to slightly condescending.
- Overly cheerful: “Oops! That wasn’t quite right — keep shining, superstar!”
- Condescending
This isn’t just a matter of style; it strikes at the core of learner motivation. When feedback feels inauthentic or makes learners feel incapable, it can undermine their sense of autonomy and competence. In contrast, a genuine, empathetic tone builds the psychological safety and trust that makes learners feel supported, not alienated, and ready to learn.
Risk #2: “Purple Prose” Language
To sound more sophisticated, AI can produce “purple prose,” or overly elaborate phrasing. It might describe a simple module as “a transformative experience that transcends mere didactic instruction.”
This leads to common issues like:
- Unnecessary metaphors
- Flowery, elaborate language
- A tendency to use the passive voice
For the learner, this inflated language isn’t just distracting; it actively works against them. It increases their cognitive load, forcing them to spend precious mental energy decoding the complex sentences instead of absorbing the core message. Conversely, clear and direct language prevents this cognitive fatigue, allowing learners to focus on understanding and retaining information.
Risk #3: Predictable “AI-Speak” Structure
Perhaps the most revealing risk is the emergence of “AI-speak.”
Ask an LLM for ten blog titles, and you might get back a list where every single one follows the exact same, overly wordy format, like “Topic: A Deeper Dive Into Subtopic.” AI models love these formulas and often rely on the same predictable sentence structures.
- “This isn’t just X… it’s Y.”
- “Whether you’re a beginner or a pro…”
- “In today’s fast-paced world…”
Individually, these phrases are fine. But when they accumulate, they create a “robotic pattern” recognizable by learners as “AI-speak,” which may cause them to disengage.
Having an authentic “human” voice with varied structure builds a memorable connection, transforming the content from a simple information transfer into a genuine interaction.
To help you spot it, here’s a quick field guide to some of the most common AI writing patterns we’ve seen in the wild.
Field Guide: Spotting AI-Speak
| The Repetitive “It’s Not X, It’s Y” Framing | “Resilience is not about bouncing back. It is about building the cultural muscle to move forward differently.” “Oversight is not about knowing all the answers — it’s about asking the right questions.” |
| The Compulsive Three-Part List | “It’s clear, concise and effective.” “Engage. Empower. Elevate.” |
| The Famous Em Dash (—) Overuse | “Early exposure — even without clear goals — primes the brain to absorb structured learning.” “Decisions like these rarely offer perfect answers — only priorities, trade-offs and long-term consequences.” |
| Formulaic Conversational Openers | “Let’s break it down.” “Let’s dive in.” |
| Overused “Corporate Speak” Words | “Catapult” “Unlock” “Drive” |
Building a “Human-in-the-Loop” Workflow
So, how do you prevent this? The key is to remember that AI is a powerful tool, but your expertise is the final voice. The most effective approach is a cycle: Start with your idea, ask AI for improvements, then edit the output with your unique human perspective.
Here are a few practical strategies to build into your workflow:
Read it aloud. Always.
This is the fastest way to catch an unnatural tone or awkward pacing. As you read, ask yourself: “Does this sound like me? Would I actually say this to a learner?”
Personalize with your perspective.
Weave in your own context, nuance, and real-world examples. Tell micro-stories to illustrate a point. This is where you add value that no AI can replicate.
Use smarter prompts.
Don’t accept the first draft. Guide the AI towards a better output with specific instructions. Try prompts like:
- “Rewrite this using a professional yet warm tone.”
- “Simplify this paragraph for a beginner audience. Avoid jargon and purple prose.”
- Get a second opinion. Ask a peer, a pilot audience, or even an AI learner persona (yeah, why not!) to proofread your writing. A fresh set of eyes can easily spot a tone that feels off or a phrase that sounds robotic.
Your Voice Is the Final Word
Learners don’t care how fast a course was built. They care how it feels. Good writing should be invisible, allowing the learner to focus on the message. When left unchecked, AI writing can become all too visible.
Use AI to handle the heavy lifting, but ensure the final product feels personal and powerfully human. If these challenges sound familiar, your team isn’t alone. Share this article with your network to start a conversation about creating more human-centered learning in the age of AI.
This article is the first step. The next step is putting these strategies into practice. What can you do to build a stronger AI skillset?
