RY

Case study · Cost of a Click

Cost of a Click

Making AI’s invisible water cost physically tangible.

An interactive installation that translates the environmental cost of AI into a physical and digital experience — connecting everyday AI interactions to real-time water and energy feedback.

Instead of asking people to understand an abstract statistic, the installation makes the consequence visible, physical, and immediate.

DePaul Summer Impact Grant 2025 · Winner

Experience the installation See how it was built
The installation at the DePaul Jarvis showcase: a lit water tank on a draped table, a monitor above it running the Cost of a Click visual, and a laptop beside it.
Role
Lead UX Researcher
Duration
10 weeks
Team
UX Researcher · UI Designer · Developer · PM
Tools
Figma · Framer · Arduino · VS Code

The outcome

We turned an invisible environmental cost into something people could see and feel.

Cost of a Click connected an AI interaction to a physical consequence: visitors used an AI system while the installation translated that activity into real-time water and digital environmental feedback.

The experience was exhibited at the DePaul Summer Showcase 2025, where exit-survey responses showed increased awareness and reported intention to use AI more mindfully.

  1. Abstract

    AI has an environmental cost.

  2. Physical

    Water moves in response to interaction.

  3. Digital

    Environmental impact becomes visible on screen.

  4. Personal

    The visitor can connect a system-level cost to their own action.

reported changed awareness of AI's environmental footprint
85%
reported intent to use AI more mindfully
73%
increase in self-reported informed decision-making
30%
TikTok views after exhibition footage circulated
6M+

DePaul Summer Impact Grant — Winner

DePaul Summer Showcase — Featured installation

How to read these: the 85%, 73% and 30% measures are self-reported participant responses. The 6M+ figure represents social reach — not product adoption or verified behaviour change.

The problem

AI feels invisible. So do its environmental costs.

AI depends on physical infrastructure — data centers, energy systems, and water used for cooling — but none of that is visible during a typical AI interaction.

The problem was not a lack of information. It was the distance between what I do → what the infrastructure consumes → what I actually perceive.

“Abstract environmental data doesn’t change behaviour. But when you watch the water pour — you feel it.”

Research finding — exit interviews after installation interaction, Summer 2025
Illustration of a lit globe wrapped in orbiting network lines.

Invisible infrastructure

Data centers · water · energy

Abstract information

Numbers · charts · reports

Everyday AI use

Ask · generate · repeat

The opportunity

Don’t explain the infrastructure. Make people experience it.

The project reframed the challenge from information delivery to experience design. The central hypothesis: a physical consequence could create a stronger connection to an otherwise abstract environmental cost.

Design principle

Sensation over statistics.

Traditional awareness

  1. Statistic
  2. Read
  3. Understand
  4. Move on

Cost of a Click

  1. AI prompt
  2. Physical response
  3. Digital feedback
  4. Reflection

Research

Research moved the project from “show the impact” to “make the impact felt.”

semi-structured interviews
15
survey responses
42
concept directions explored
3
usability test rounds before exhibition
2
  1. Desk research
  2. 15 interviews + survey n=42
  3. Journey mapping + synthesis
  4. 3 concept directions
  5. Physical + digital prototype
  6. 2 rounds of testing

The insight

Abstract environmental data was too easy to ignore.

We had to collapse the distance between an AI interaction and the infrastructure supporting it.

That led to the defining experience decision: give the environmental cost a physical form.

One AI interaction

Real-time water response

Digital environmental feedback

Felt cost

The experience

Every interaction creates a visible consequence.

The physical and digital layers were deliberately coupled, so the environmental cost never arrived as a separate educational message alongside the interaction — it was the interaction.

  1. 01

    Ask

    The visitor enters or submits an AI query.

  2. 02

    Detect

    The system registers the interaction.

  3. 03

    Translate

    The interaction is converted into a physical response.

  4. 04

    Experience

    Water moves while the digital interface updates.

  5. 05

    Reflect

    The visitor sees the environmental cost behind the interaction.

The installation running: the lit tank, the monitor above it, and the laptop visitors typed on.
Ask — the laptop and tank as a visitor meets them.
The submersible pump sitting in a bucket of water, tubing and power cable running out of frame.
Physical response — the pump that moves the water.
The four-panel video wall showing the Cost of a Click title card and a QR code reading “Scan for more information”.
Environmental data — the wall that carried the context.

System design

From prompt to pump.

One interaction drove two loops at once. The physical layer made the cost tangible. The digital layer explained what that physical response meant.

AI interaction

Physical loop

  1. Arduino
  2. Relay / sensor logic
  3. Water pump
  4. Physical water response

Digital loop

  1. Digital interface
  2. Real-time environmental feedback
The Arduino board wired to a relay module, with the pump's power line running through it.
The Arduino and relay that switch the pump on each interaction.
The pump seated in its reservoir of water.
The pump in its reservoir — the other end of the same circuit.

Design rationale

Water gave an invisible system a physical volume.

A number can communicate scale. A physical quantity can communicate consequence.

The installation used real water as a tangible representation of infrastructure people normally never see.

  1. AI query
  2. Invisible water use
  3. Physical water response
  4. Visible consequence

Secondary metaphor

The digital environment showed what the physical interaction could not.

As visitors interacted, the digital environment added a second layer of consequence through visual change: a tree withering over time to represent the accumulated impact of repeated AI use.

Physical feedback creates immediacy. Digital feedback creates narrative.

  1. Healthy

  2. Stressed

  3. Withering

Diagram of the on-screen progression, not a screenshot.
The project website shown on a laptop: “Invisible Impact. Visible Change.”, a DePaul University Impact Grant 2025 badge, and figures for weekly active users, daily queries and gallons of water.
The project website, which carried the research beyond the room.

My role

I connected research, experience design, and physical interaction.

As Lead UX Researcher I led the research process and worked across experience design, sensor interaction, and testing.

The role extended beyond interface design: I translated findings into the physical and digital behaviour of the installation, and helped shape the experience from concept through exhibition.

  1. Research

    • Desk research
    • 15 interviews
    • Survey n=42
    • Journey mapping

  2. Experience design

    • Problem framing
    • Concept directions
    • Interaction model
    • Information hierarchy

  3. Physical-digital design

    • Water interaction
    • Sensor behaviour
    • Digital feedback
    • Arduino integration

  4. Validation

    • Lab usability test
    • Field test
    • Iteration

  5. Exhibition

    • Showcase prep
    • Audience interaction
    • Impact measurement

The process

Research-heavy at the start. Test-heavy at the end.

  1. 01Discovery

    Weeks 1–2

    Desk research · secondary synthesis · problem framing

  2. 02User research

    Weeks 3–4

    15 interviews · survey n=42 · journey mapping

  3. 03Concept + design

    Weeks 5–6

    3 directions · prototyping · Figma flows

  4. 04Build

    Week 7

    Arduino · sensor integration · water pump

  5. 05Test R1

    Week 8

    Lab usability test · iteration

  6. 06Test R2

    Week 9

    Field test · refinement

  7. 07Exhibit

    Week 10

    DePaul Summer Showcase · impact measurement

Iteration

Testing exposed problems the concept couldn’t.

Two rounds of usability testing before the exhibition surfaced physical and interaction friction.

The experience had to work as a physical system, not just as a concept.

  1. Issue

    Water spillage

    Fix

    Refined the physical setup.

  2. Issue

    Prompt timing

    Fix

    Adjusted the interaction timing.

  3. Issue

    Screen readability

    Fix

    Improved the digital hierarchy.

Two teammates assembling the reservoir: one lowering the pump into the water, the other watching from a chair.
Assembling the reservoir and seating the pump.
Four team members standing beside the finished installation — the tank, the monitor and the Cost of a Click poster.
The installation assembled and running, before doors opened.

Behind the build

The prototype had a body, a screen, and a feedback loop.

  • Physical

    Water container · pump · tubing · physical feedback

  • Compute

    Arduino · relay · sensor logic

  • Digital

    AI interaction · environmental counter · data visualization

  • Experience

    User action · immediate consequence · reflection

Exhibition

The final test was putting the experience in front of people.

Cost of a Click was exhibited at the DePaul Summer Showcase 2025, where students, faculty and public visitors experienced the installation directly.

The exhibition room: the lit tank on the DePaul-draped table on the left, the four-panel video wall on the right.
The room as visitors found it — installation on one side, video wall on the other.
The four-panel video wall with the Cost of a Click title card and a QR code.
The video wall, with a QR code to the project site.
Printed take-aways laid out on a table: brochures, bookmarks and hand fans in the Cost of a Click pattern.
Informational material designed to travel home with visitors.
The full team standing either side of the installation table at the DePaul Summer Showcase.
The team at the DePaul Summer Showcase, 2025.

Impact

The installation made the invisible visible — and the response was measurable.

The result wasn’t simply that visitors saw a new data visualization. The experience created a stronger connection between AI use and the infrastructure behind it.

reported changed awareness of AI's environmental footprint
85%
reported intent to use AI more mindfully — specifically reducing unnecessary or casual queries
73%
increase in self-reported informed decision-making about when to use AI tools
30%

Qualifier: these outcomes are based on self-reported exit-survey responses. They should not be presented as proof of long-term behaviour change.

Beyond the exhibition

The conversation left the room.

After the DePaul exhibition, footage of Cost of a Click circulated on TikTok and reached more than 6 million views, broadening the public conversation around AI’s hidden environmental cost.

6M+ views measure reach — not adoption, and not behaviour change.

Originally posted by @thetshegofatso on TikTok · 6M+ views · 2025

Concept — where this could go next

Illustration, not the exhibited installation

The showcase build was one table, one tank, one pump. These illustrations imagine the same idea at exhibition scale — a room a visitor walks into rather than a station they walk up to.

Illustration of a gallery-scale version: a suspended cluster of lit screens, with wall text reading “Ask. Explore. See the impact.”
Concept illustration — the installation imagined at room scale. Not a photograph of the showcase.
Illustration of a visitor at a lit kiosk reading “Ask. Explore. See the impact.” with a tap-to-begin prompt.
Concept illustration — a walk-up entry point to the same interaction.

Capabilities

This wasn’t just a sustainability installation.

  • UX research

    Interviews · survey · synthesis · testing

  • Experience design

    Interaction model · information hierarchy · physical-digital journey

  • Systems thinking

    AI interaction → sensor → physical response → digital feedback

  • Prototyping

    Figma · Framer · Arduino · VS Code

  • Physical computing

    Pump · relay · sensor interaction

  • Public-facing design

    Exhibition · audience testing · impact measurement

Key takeaways

Design the experience, not just the information.

  • Felt information is different

    Physical feedback created a different kind of environmental awareness than a static number.

  • Research can shape hardware

    Research influenced the physical interaction, not just the screen design.

  • Physical systems need different testing

    Water spillage, timing, spatial setup and readability all became UX problems.

  • Ethical framing matters

    Inform without relying on fear, and leave people with something actionable.

Continuing research

The installation became a research platform.

Further work on the project developed into a paper on interactive art as a public-facing AI ethics interface.

  1. Installation
  2. Research evidence
  3. Academic paper

Cunningham, J. L., Iqbal, S., Nacu, D., Caplan, B., Yerramilli, R., & Mukkamala, D. D. (2026). “Making AI Infrastructure Visible: Interactive Art as a Public-Facing AI Ethics Interface.” AAAI/ACM Conference on AI, Ethics, and Society (AIES 2026).

Paper #343 · Under review — not accepted

Final thoughts

The invisible has a cost. Let’s make it visible.

Cost of a Click started with a communication problem: people could understand AI’s environmental impact without ever feeling connected to it.

The project turned that abstract cost into an experience — something people could see, touch, question, and remember.

Experience the project Back to projects

Up next

About me

Read more