Everyone can now generate images and videos. Nearly for free. But did you see this year's April Fools content? As Wharton's Ethan Mollick put it, the tools are the most powerful in history, yet the output is still mediocre.
What's going on?
Mollick's April 1st observation was simple. In 2026, anyone can generate almost any image or video they can think of for nearly free — yet April Fools posts were basically just as bad as any other year. This isn't just a joke. It's a signal that execution capacity is now infinite while imagination stays flat.
Research keeps piling up to support this. A massive University of Montreal study found that AI surpassed average human creativity across 100,000+ participants, but still fell short of the top 10% most creative humans — especially in deeper creative work like poetry and storytelling.
Wharton's Terwiesch team found something even more specific: using ChatGPT for ideation improves individual idea quality but dramatically reduces idea diversity. In one experiment, only 6% of AI-assisted ideas were unique, compared to 100% in the human-only group.
Exeter University's Ahmed-Kristensen study reached the same conclusion. AI excels at idea volume (fluency) but 600 average people crushed AI on idea differentiation (flexibility). The takeaway: what AI can't do isn't "produce a lot" — it's "think differently."
What's actually changed?
Execution used to be the bottleneck. Great idea but can't design, can't film, can't code? Game over. Now it's the opposite. AI handles execution, but imagining what to create has become the hardest part.
| Pre-2023 | 2026 Now | |
|---|---|---|
| Bottleneck | Execution (design, code, video) | Imagination (what to build) |
| Scarce resource | Technical skill | Original ideas |
| AI's role | Assistant tool | Execution engine |
| Competitive edge | Making it well | Thinking differently |
| Content problem | Not enough | AI slop flood |
UF marketing professor Zou's research puts numbers to this shift. As AI-generated content floods platforms, recommendation systems get congested, making it harder to surface genuinely good content. The more volume there is, the more quality differentiation matters — a paradox.
Mollick frames this through his "Jagged Frontier" concept. AI capabilities don't grow evenly. Image generation, coding, and analysis are superhuman, but memory and genuine novelty remain weak spots. Those jagged edges are exactly where humans need to fill in.
Watch out for the "Imagination Paradox"
The moment AI gives you 12 options, your brain switches from generative mode to evaluative mode. These two modes neurologically inhibit each other. Seeing AI options first shuts down your original thinking.
How to unclog the creativity bottleneck
- Imagine before you prompt
Write down at least 3 directions before touching AI. The weirder and riskier, the better. Prompting AI first switches off your divergent thinking. - Use AI as a mirror
Compare your ideas with AI output. If they match closely, you're trapped in the statistical average. The wider the gap, the more original you are. - Diversify your prompts
Ask the same question in different ways. The Wharton team's "chain-of-thought prompting" — breaking tasks into structured steps — increases idea diversity. - Mix multiple models
Throw the same question at ChatGPT, Claude, and Gemini. Different models pull from different distributions, widening your idea range. - Create deliberate friction
Add constraints on purpose. A 280-character tweet, a 3-color design, a 5-word slogan — constraints force creativity. Mollick himself emphasizes that style matters more than ever in the AI era.
The 80/20 Imagination Rule
Spend 80% of your creative energy imagining independently without AI. Use AI for only 20% — refinement and exploration. Imagination is a muscle: use it or lose it.
