A
B
C
D
E
100

This natural resource is used to cool AI data centers.

Water

100

Cooling prevents these machines from overheating.

Servers or data centers


100

Training this type of advanced technology can use millions of liters of water.

Artificial Intelligence (AI)

100

AI companies build these to power and cool their models.

Data centers and servers

100

Water usage by AI impacts this natural system.

Environment or water cycle

200

Cooling systems often rely on this method, which evaporates water to remove heat.

Evaporative cooling

200

This sustainable practice reuses water in cooling systems.

Water recycling

200

This large AI model developed by OpenAI required lots of energy and water.

ChatGPT

200

High water use in AI can lead to shortages where

Everywhere

200

If a server gets too hot and water cooling fails, it may do this.

Shut down or crash

300

Using too much water in dry areas can contribute to this environmental problem.

Drought
300

This measurement tells us how much water is used for a specific task.

Water footprint

300

In addition to water, data centers use a lot of this resource.

Electricity

300

This type of AI use (common in apps like ChatGPT) still consumes water even after training.

Inference or running AI 

300

A data center that uses solar panels and water recycling is called this.

Sustainable data center or Green data center


400

Name one alternative to water-based cooling.

Air, immersion or geothermal cooling

400

This tech giant paused data center plans due to water concerns in the Netherlands.

Microsoft

400

AI water use is mostly invisible to users and is called this kind of cost.

Hidden or indirect cost

400

Training bigger models increases water use for this reason.

More to compute = more heat = more cooling

400

Countries facing water stress may oppose this type of AI expansion.

AI infustrcture

500

This role in tech companies is responsible for tracking environmental impact.

Sustainability officer or environmental engineer

500

AI companies might be asked to publicly share this kind of water usage data.

Water footprint reports

500

Give one solution to reduce AIs water footprint without stopping innovation.

Renewable energy, building in cooler climates, optimizing models, as people we can change the way we prompt

500

This term refers to making sure AI helps people without harming the planet.

Ethical or responsible AI