LINDA BROWN
WorkGame Based Learning
AI SystemsEducationGames
GAME BASED
LEARNING

When AI visuals shape how children experience learning.

As generative AI made visual media possible at scale, the challenge became not whether we could generate visuals — but how to do so responsibly for young learners.

At A Glance
RolePrincipal Product Designer, AI Systems & Media
PartnersLearning Science, Engineering, Content
TimelineMulti-phase initiative
AudienceK–5 learners and educators
FocusAI media pipeline, character systems, visual consistency
The Real Challenge

Scaling creativity without sacrificing care.

Educational media for children must meet a higher standard — emotionally safe, developmentally appropriate, inclusive, and aligned with learning goals.

Key Challenges
01

Preventing visual inconsistency across AI-generated assets at scale

02

Avoiding bias, stereotypes, or inappropriate representations

03

Maintaining a cohesive visual language across grades and regions

04

Supporting multiple literacy levels without adding visual distraction

My Role & Scope

Designing the system behind the visuals.

Led the design of the AI media system — defining how characters, environments, and scenes were generated, constrained, reviewed, and scaled. Accountable for both the creative output and the governance that made it reproducible.

Direct Ownership

AI prompt architecture and generation constraints

Character and environment design frameworks

Visual consistency standards across the product ecosystem

Strategic Influence

Ethical AI usage guidelines for young learners

DEI considerations embedded into generative media prompts

Alignment between visual outputs and instructional goals

Designing the AI Media System

From prompt to picture — with intention at every step.

I designed the media pipeline itself — ensuring every visual output was shaped by clear rules, values, and review loops that couldn't be bypassed.

01

Encode Intent

Instructional & emotional goals embedded into prompt architecture

02

Constrain

Safe visual bounds applied — style, diversity, complexity rules

03

Generate

AI creates assets within governed parameters

04

Review

Human-in-the-loop checkpoints before any asset ships

05

Scale

Approved patterns replicated across grades, regions, and teams

The solution included

Prompt guidebooks and reusable templates for distributed teams

Style and character constraints that bounded AI generation

Environment logic tied to narrative tone and literacy level

Review and iteration checkpoints before any asset was approved

Designing Characters & Worlds

Consistency, warmth, and clarity for early learners.

Characters and environments were designed to feel welcoming and emotionally safe — especially for students still building reading confidence.

Gentle facial expressions and body language that signal safety, not pressure

Clear silhouettes and visual hierarchy to reduce cognitive load

Calm, cohesive colour palettes that support focus without monotony

Deliberate avoidance of visual noise or overstimulation

Character — supportive
Character — curious
Environment — forest
Environment — classroom
Scene — adventure
Scene — calm
Key Decisions & Guardrails

Where restraint mattered more than possibility.

Consistency over infinite variation

Familiarity supports comprehension. We deliberately limited variation so children could build reliable mental models of the visual world.

Ethics embedded into prompts, not policies

Guardrails were proactive, not reactive — ethical constraints lived inside the prompt architecture before generation happened, not after it failed.

Limited visual complexity, intentionally

Every added visual element had to earn its place by supporting learning, not just appealing to attention.

Designed for teams, not just assets

Documentation and training were core deliverables. A system only I could operate wasn't a system — it was a bottleneck.

How This Media Shows Up in the Product

Supporting learning without distraction.

Within the learning experience, AI-generated visuals supported comprehension and motivation without becoming the focus themselves.

Reinforce understanding
Reduce intimidation
Support independent progress
Create moments of delight without pressure
Outcomes & Signals

What this system enabled.

Teams could produce high-quality, consistent visuals at scale — across grades and regions — without reinventing the rules each time.

Ethical and instructional alignment was maintained throughout production without slowing creative output or requiring constant review bottlenecks.

Manual asset creation time dropped significantly through reusable prompt frameworks, templates, and pre-approved visual patterns.

Educators reported stronger student engagement and comfort — particularly among early or struggling readers encountering the characters.

Reflection

Designing AI with care.

This project reinforced that AI design is not about capability alone — it's about responsibility. The hardest design decisions weren't technical; they were ethical.

“AI design is not about capability alone — it's about responsibility.”
Related Work
← PreviousSmart Grader
Next →Outryd