Preference And Effectiveness Are Not The Same
I like pies. My preferred way of consuming data is through pie charts.
Nobody says this. At least, I haven't heard much of it as a data professional. If you said something like that in a meeting, you'd get the polite laugh reserved for people who might be joking. Because we all understand, instinctively or by trade, that liking a thing has nothing to do with whether that thing helps you understand anything or do anything better. Or, to be more precise, the message or story you're sharing determines the most effective way to communicate it to your audience. Getting something you like can increase satisfaction and motivation. But that's different than learning.
A pie chart of 14 nearly equal slices tells you nothing. A pie chart comparing values over time tells you less than nothing (it actively misleads). And no amount of personal fondness for pastry changes what the human eye can and cannot judge. We're bad at comparing angles. We're good at comparing lengths. Let alone angry 3D charts!!
That's why the boring bar chart keeps winning. Your preference doesn't get a vote. We, data professionals, pick the visualization based on the data. Whether you'd prefer to see more pies or not, that's totally your call.
But I enjoy pie charts more.
Sure. You might also enjoy reading quarterly financials as a limerick. The question was never what you enjoy. The question is what helps you see the data insights.
The Pie Charts Of Learning
Which leads me to the never-ending story of learning and learning styles. We've spent decades doing exactly this with a straight face: someone declares "I'm a visual learner," and instead of the polite laugh, they get a redesigned course with lots of graphics. We survey people about their preferred learning style. We sort them into buckets: visual, auditory, kinesthetic. We build content to match. We call it learner-centered design and feel good about it.
When researchers went looking for evidence that matching instruction to learning styles improves learning, they didn't find weak evidence. They found essentially none, and the few properly designed studies contradicted the idea (Pashler, McDaniel, Rohrer, and Bjork, 2008). Yet a systematic review found that roughly 89% of educators still believe in matching instruction to learning styles, and most report actually doing it (Newton and Salvi, 2020). And now AI is trained on all that myth. No wonder AI agents are as confused as humans.
The Confusion
We confuse two different questions. "What do people prefer?" is a real question. It's worth asking. Preference affects motivation, and motivation matters. "What actually works?" is a different question. It has a different answer. And when the two conflict (as they often do), "what works" has to win, because the learner's goal was never to be catered to. It was to get better at something.
The person who wants everything in pie charts doesn't need more pie charts. They need someone to show them a bar chart and say: Look how much faster you just found the answer. The self-declared visual learner doesn't need their safety training converted to infographics. They need retrieval practice, spacing, feedback, worked examples, the unglamorous things with actual evidence behind them, regardless of which sensory channel they claim as their brand.
Measurement Is Key
Preference is data. But it's data about comfort, not effectiveness. Design to it, and you optimize for how learning feels instead of whether it happened. Those two measures diverge more often than we'd like. Smooth, enjoyable experiences routinely produce worse retention than effortful, slightly uncomfortable ones. If you've ever seen glowing course evaluations sitting next to flat performance numbers, you've watched the divergence happen.
So the next time someone asks for the training version of a pie chart: shorter, prettier, matched to their style, take the request seriously as a signal about motivation. Then ask the better question. Not "what do you like?"
"What would it take for you to be measurably better at this in 90 days?" And measure it. Nobody has ever answered that one with "more pie."
Wait, It Gets Worse With AI
For decades, one thing quietly protected us from our own bad theory: cost. Cost of resources, time, and effort to produce "learning style" matching content. Building three versions of a course (visual, auditory, and kinesthetic), just 3 out of the 75 different styles, was expensive. Budgets forced trade-offs, trade-offs forced questions, and somewhere in the process someone usually asked, "wait, do we actually need this?" Friction was our accidental quality control.
That friction is gone. Get ready for individualized and personalized pie chart courses.
AI can now generate a version of your course for every learner. Not 3 styles, all 3000. A podcast version for the "auditory learner." An infographic for the "visual learner." A simulation for whoever checked "hands-on." Each one is produced in minutes, each one is polished, each one is personalized to the point that humans can't even keep up with the pace. The dashboards will glow. Learners will report loving it. Salespeople will get funny AI stories they can play at double speed about product knowledge. Win-win!
- Did I mention measurement matters?
From now on, it will be the only thing that matters. If the underlying theory AI was trained on is wrong, we haven't actually produced personalized learning. We've industrialized the pie chart. In 3D with rainbow colors shading that talks.
AI doesn't validate our design assumptions. It amplifies them. You feed it a myth, and it will execute that myth flawlessly, at scale, with a confidence that looks a lot like evidence. A bad idea used to fail slowly, in one course, where someone might notice. Now it can fail beautifully across an entire enterprise, wrapped in the word "adaptive."
There's a second trap hiding inside the first: measuring the wrong thing. When AI is trained to satisfy learners by giving them what they want (rather than what they need), satisfaction scores will skyrocket. Smooth, engaging, and entertaining. Feels good to learn. The problem is that the struggle is what makes practice stick. The unfamiliar format that forces real attention. The wrong answer you have to sit with before the reveal. An optimization loop pointed at satisfaction will happily optimize the learning right out of the learning.
None of this makes AI the villain. The same machine that can generate infinite pie is the machine that can finally do the things we never had capacity for: adapt to what a learner knows rather than what they like, generate retrieval practice at the exact edge of someone's competence, space it over weeks, vary the format deliberately not to match a style, but to break the comfort of one.
We, humans, are responsible for how we implement AI for learning. We, humans, are responsible for selecting and influencing tech vendors' "AI solutions" approach. We, humans, are responsible for saying no to courses when they are not needed, and yes to solutions that may fall outside of the traditional expertise. We don't need to write white papers to convert learning-style believers. We need to put a measurement in place that shows the impact on the job. Not just memorization or recall after a program, but real, sustained behavior change that happens under realistic circumstances.
"Feed the content" is as horrible as 3D pie charts for all data. The problem is not AI. The problem is human. We have the opportunity to stop hiding behind a lack of resources and technology to do the right thing. No more excuses. But, without systems thinking and focus on behavior change that drives impact, this "feed the content" might just turn into what it says: upload your PDF, and it magically creates pie charts from it.
No pies were hurt during this article.
References:
- Pashler, H., M. McDaniel, D. Rohrer, and R. Bjork. 2008. "Learning styles: Concepts and evidence." Psychological Science in the Public Interest 9 (3): 105–19.
- Newton, P. M., and A. Salvi. 2020. "How common is belief in the learning styles neuromyth, and does it matter? A pragmatic systematic review." Frontiers in Education, 5, 602451.
