Pattern Recognition → Utilisation → Creation. This isn't a framework we invented. It's the progression that every major learning model has validated for decades — applied to AI.
All skill acquisition follows the same arc: you first learn to recognize patterns, then you learn to use them, then you learn to create new ones.
This has been proven in chess (Chase & Simon, 1973), medicine (Dreyfus, 1980), and cognitive science (Anderson's ACT-R). The same progression applies to AI: recognizing where AI fits, using it effectively, and building systems that scale.
The GBI challenge uses this progression. Each 10-day level begins building one of these capabilities through evidence-based micro-learning and spaced practice.
You learn to see where AI fits and where it doesn't. You develop judgment — recognizing good output from bad, the right tool from the wrong one, an AI-shaped task from a human-shaped task.
Research: The Harvard/BCG "Jagged Frontier" study (2023) found that people who recognized which tasks fit AI outperformed by 40%. Those who couldn't recognize the pattern performed worse with AI than without it.
You move from recognition to application. AI becomes part of how you actually work — not a novelty you try once, but a tool you iterate with, refine through, and integrate into real workflows.
Research: Anderson's ACT-R theory calls this the "procedural stage" — where declarative knowledge (knowing about AI) becomes procedural knowledge (working with AI). PwC's data shows this fluency carries a 56% wage premium.
You create new patterns. AI agents, automations, pipelines that work without you. You're not just using AI as a tool — you're designing systems that scale independently.
Research: This maps to the "adaptive" stage in ACT-R and "Extended Abstract" in the SOLO Taxonomy — where you generalize learned patterns to entirely new contexts. WEF ranks creative thinking as the #2 skill for 2025-2030.
Let's be honest: 10 days of 5-minute practice won't make you an expert. Deep pattern recognition takes thousands of hours (Chase & Simon, 1973). Habit automaticity takes a median of 66 days (Lally et al., 2010).
But here's what the science says about what 10 days can do:
The steepest part of the curve
The power law of learning proves that early repetitions produce the largest gains. Day 1-10 is where the most progress per minute happens.
Micro-learning works
5-minute daily sessions improve retention by 25-60% compared to traditional learning formats. Frequency matters more than duration.
Blocked practice for beginners
Our 3-habit rotation (3 days each) is "blocked practice" — proven more effective for novices than interleaving. Three consecutive days gives 2 overnight sleep consolidation cycles.
Bounded challenges sustain motivation
Open-ended commitments fail. Bounded challenges (7-10 days) maintain engagement. Zech's own 30-day challenge saw major drop-off — even he quit at day 20.
Identity shift begins immediately
James Clear's research shows that even a few days of "being the kind of person who practices AI" can shift self-concept. The badge isn't just proof — it's identity.
We didn't invent this progression. Every major learning and AI competency framework independently arrived at the same three stages:
| Framework | Recognition | Utilisation | Creation |
|---|---|---|---|
| Bloom's Taxonomy(2001) | Understand / Analyze | Apply | Create |
| Anderson's ACT-R(1982) | Declarative (know it) | Procedural (do it) | Adaptive (create with it) |
| SOLO Taxonomy(1982) | Identify patterns | Integrate patterns | Generalize to new contexts |
| Dreyfus Model(1980) | Rule-following | Competent application | Expert intuition |
| UNESCO AI Framework(2024) | Understand | Apply | Create |
| Microsoft AI(2026) | Fundamentals | Use tools | Build agents |
| Google AI(2025) | AI Essentials | Prompt & apply | Domain fluency |
| Mollick (Wharton)(2024) | Exposure | Calibration | Integration / Mastery |
What's proven
What we believe
What we don't claim
Anderson & Krathwohl (2001) — A Taxonomy for Learning, Teaching, and Assessing — revised Bloom's Taxonomy
Anderson, J.R. (1982) — Acquisition of cognitive skill — ACT-R theory of declarative → procedural → adaptive knowledge
Biggs & Collis (1982) — SOLO Taxonomy — Structure of Observed Learning Outcome
Dreyfus & Dreyfus (1980) — A Five-Stage Model of the Mental Activities Involved in Directed Skill Acquisition
Chase & Simon (1973) — Perception in Chess — pattern recognition as the foundation of expertise
Dell'Acqua et al. (2023) — Navigating the Jagged Technological Frontier — Harvard/BCG study of 758 consultants using GPT-4
Mollick, E. (2024) — Co-Intelligence: Living and Working with AI — Wharton School
Lally et al. (2010) — How are habits formed — 66-day median to automaticity, asymptotic early gains
World Economic Forum (2025) — Future of Jobs Report — analytical thinking #1, creative thinking #2 skill for 2025-2030
PwC (2025) — AI Jobs Barometer — 56% wage premium for AI-fluent professionals
McKinsey (2024) — AI adoption research — 25% faster task completion with AI workflows
UNESCO (2024) — AI Competency Framework for Students — Understand → Apply → Create
The steepest part of the learning curve is the first 10 days.
Take the AI-Ready Challenge