There are those moments in the daily L&D routine that feel like a time warp. You open an authoring tool on Monday morning, type a few lines, and before your coffee has reached its ideal temperature, there it is: a script. Then a quiz. And while you’re at it: an explainer video with an avatar, please in a ‘friendly-competent’ tone, 90 seconds, plus two versions for different target audiences.
Learning content used to be a craft. Today, it’s an assembly line, not because we suddenly value quality less, but because production capacity has exploded. This very shift – from content as a scarce commodity to industrial availability – is analyzed by Dr. Philippa Hardman in her text “The AI Content Explosion: What Your Learners Actually Think (And Why It Matters)“ (December 11, 2025).
Our colleague Maria Matthäus has condensed the core statements from it internally in such a way that you want to nod, grin, and briefly swallow all at once while reading.
Hardman’s perspective is refreshingly unromantic: The exciting question is no longer whether we use AI-generated content, but rather: What does that do to the people who are supposed to consume this content? And – even more importantly – how does that feel for learners? Because in a learning context, “feeling” is not decoration. It is often the driving force.
What Learners Will Actually Find at the End of 2025 – The New Normal
Reading Hardman’s synthesis (study situation 2023–2025), it seems almost trivial: learners have long encountered AI not just as a chat window. AI has now become an entire media family.
For example, there are:
- AI tutors & chatbots that respond around the clock and remain friendly, no matter how often you ask the same question.
- Synthetic instructors / AI videos that are increasingly replacing classic “talking-head” videos.
- AI-generated assessments (question pools, adaptive quizzes, practice tasks) that impress in quantity until the first error is noticed.
- Personalized learning paths that adapt to preferences and performance, thereby quietly shaping learning biographies.
- Multimodal content (text, audio, image, video in combination) that either brilliantly relieves burden or overwhelms cognitive capacity, depending on how well (or poorly) it is designed.
The crucial point: In the debate, we – as an industry – conspicuously like to talk about production efficiency. Learners, however, experience learning reality. And that is not automatically congruent with “produced faster.”
The Surprisingly Nuanced Attitude of Learners
Hardman’s findings initially sound reassuring: learners are often cautiously optimistic. Many find AI convenient, accessible, and useful. But: this approval is selective. It depends on the content type, the context, and something very human: trust.
1) Chatbots: loved – and simultaneously misunderstood as an “answer machine”
When learners talk about AI tutors, it often sounds like an ode to availability: immediate answers, no waiting for office hours, no embarrassment about “stupid questions.”
And then comes the second sentence: Many use bots not as thinking partners, but as a shortcut. “Give me the solution.” Period. That’s human. And didactically explosive, because it can weaken self-regulated learning, which is precisely what we actually want to strengthen.
2) AI Videos: accepted, but not welcome everywhere
For procedural topics (step-by-step, demos, basics), AI videos can certainly keep up with understanding, according to Hardman’s overview, if the script and didactics are correct.
However, as soon as it comes to relationships, ethics, identity, or emotionally charged topics, the preference clearly shifts towards humans. And then there’s that special case of modern media psychology: the Uncanny Valley effect. Hyper-realistic avatars that are “almost” human don’t appear futuristic to many learners, but… strange. Or unsettling.
This is not an aesthetic side issue. Discomfort reduces social presence, and social presence influences whether learners ask questions, doubt, or persevere.
3) AI Assessments: “more” is good – until “wrong” appears
For practice questions and quizzes, learners like the quantity: more material, more variations, more training. But AI items have a credibility problem as soon as ambiguous question stems, incorrect answer keys, or factual errors occur. And that’s the point where a “cool AI course” quickly becomes “this course is sloppy,” even if the rest is solid.
4) AI Examples & Sample Solutions: helpful – but timing determines the effect
Here, Hardman’s text becomes almost literary, as it describes a simple pattern: learners love examples because examples provide security. But security in learning is a mixed blessing.
If examples come too early, it becomes comfortable, and comfort is often the enemy of competence. If examples come after an initial attempt, this productive friction arises: comparing, reflecting, correcting. And precisely this sequence appears in studies to lead to better later performance without AI.
The Central Paradox: Satisfaction Does Not Equal Ability
Hardman formulates a sentence that sticks because it is uncomfortably plausible:
“They feel great while learning less.”
This is not cultural pessimism. This is an indication that learner experience (confidence, satisfaction, “feels good”) and learning outcome (competence, transfer, independent performance) can diverge, especially for beginners.
And this is where it gets exciting for L&D, because it highlights an old truth anew: people evaluate learning offerings not according to the didactic blueprint, but according to what they feel. Trust, social presence, the feeling of “this is worth my effort” – all of this can influence engagement more strongly than technical brilliance.
Why This Should Matter to Us – Even If We Are Not Tech Skeptics
The most convenient misconception of the last two years was: “If it’s produced faster and people like it, it’ll be fine.” Hardman’s overview shows: It’s not that simple. AI content can indeed work, but learners don’t react to “AI” as a label, but to suitability, transparency, accuracy, and human connection.
And with that, we are already transitioning to the next chapter: While we are still discussing whether an avatar smiles “sympathetically enough,” AI is just beginning to do entirely different things.
Because the next wave is not called “Content.” It’s called: Agents.
(To be continued in Blog Post 2: What happens when AI not only creates content – but sorts our files, writes reports, and lives on the desktop as a “colleague”?)
