Listen to the Podcast Episode for a deeper dive

Small Sustainable AI Choices We Can Control | AI for Interior Designers™

AI for Interior Designers™ Podcast

Small Sustainable AI Choices We Can Control

Episode 50 — a milestone worth marking with a topic most AI conversations skip: what AI actually costs the environment, what it does not, and the small choices designers have right now that matter far more than giving up their tools.

Ep. 50 Solo Episode Episode 50 🎉 ✦ Jenna Gaidusek May 2025

This blog was written using AI as a recap from the recording, then edited by the author for accuracy and details.

Key Takeaways

Using AI 10 times a day for an entire year generates about 16 kg of CO₂ — the equivalent of running your dryer 16 times or taking eight hot showers. Not nothing, but genuinely small compared to other daily habits.
Switching your home's bulbs to LEDs saves 433 kg of CO₂ per year — 27 times the annual emissions of daily AI use. Small, accessible choices at home vastly outweigh the environmental cost of your AI workflow.
Data center water consumption is the more pressing concern — especially in water-stressed regions where data centers draw from municipal drinking water supplies. Only 16% of major data center companies publicly disclose their water management plans.
AI can help designers be more sustainable, not just less. Faster sourcing, better material research, reduced order errors, and layout accuracy improvements all reduce waste and resource consumption in the projects you design.
Designers are positioned to lead on sustainability in a way most industries are not. Every material specification, lighting recommendation, and plumbing choice is an opportunity to build a more sustainable world — and to model that for clients who may not yet know the difference.

What AI Actually Uses — The Numbers in Context

The environmental conversation around AI is loud, but it frequently lacks the context that would make it actionable. Jenna runs the numbers using Perplexity-sourced research — and the comparison is genuinely surprising.

Daily AI use

16 kg

CO₂ per year from 10 LLM searches per day — every day for 12 months

Switch to LED bulbs
433 kg
CO₂ saved per year — 27× the annual emissions of daily AI use

The equivalence comparisons make this concrete. Your annual AI use at 10 searches per day is roughly equal to:

🚗

Driving a gas car 39 miles

Skipping one unnecessary mile of driving offsets 3 weeks of daily LLM searches.

🌀

Running your dryer 16 times

You could use ChatGPT 10 times a day for an entire year — or run your dryer 16 times. Same emissions.

🚿

Taking 8 hot showers

One 10-minute hot shower emits ~2 kg CO₂. Your entire year of AI use equals 8 showers.

💡

Leaving a 100W bulb on for 93 days

Leaving a single standard bulb on 4 hours a day for just over 3 months equals your entire annual AI footprint.

The point is not to dismiss AI's environmental cost — it is real. The point is to make the trade-off legible. The lifestyle and design choices within a designer's direct control have far more environmental leverage than whether or not they use an LLM for sourcing research.

The Water Problem — Which Is Bigger and Less Discussed

While the carbon conversation around AI is relatively well-covered, the water consumption story is more alarming and less regulated. Data centers cool their computing infrastructure using enormous amounts of water — in most cases, drawing from municipal supplies that also serve residential homes and agriculture.

The scale: a single large data center can use up to 5 million gallons of water per day. In 2023, Google's data centers used more than 6 billion gallons of water globally. In water-stressed regions like Phoenix and other parts of the American Southwest, data centers collectively consume over 170 million gallons daily — in areas where residential water restrictions already exist.

Most of the traditional cooling method — evaporative cooling — evaporates the water entirely. It does not return to the supply it came from. The water is gone, and the supply must be replenished from the same municipal sources that serve the surrounding community.

"They are taking the water out of people's homes and putting it into AI cooling systems because it's cheaper — and because nobody told them no."

— Jenna Gaidusek

The transparency problem is significant: only 16% of major data center companies publicly disclose their water management plans, according to an NPR report. Many municipalities have not historically required water balance studies or sustainable measures as a condition for permitting new data centers. Water for industrial users is priced cheaply and not tied to scarcity — which means the economic incentive is entirely in favor of the status quo.

There are better options: closed-loop liquid cooling systems that circulate coolant in a sealed loop with little to no water loss, and reclaimed water solutions that use treated wastewater rather than potable drinking water. Microsoft has committed to reducing water use in specific data centers by 95% and becoming water-positive by 2030. California is introducing legislation requiring data centers to track and report water and energy usage. These changes are happening — but not as fast as the infrastructure is expanding.

What Designers Can Actually Do — High-Impact Choices

The good news is straightforward: a small number of common design specifications and home choices have dramatically more environmental impact than AI use. These are all within a designer's sphere of influence — either in their own studio and home, or in client projects.

Switch to LED Lighting

The single highest-impact residential change. Applicable in every client project.

Saves ~433 kg CO₂/year — 27× annual AI footprint

Smart Thermostat

Upfront ~$350. Cuts heating and cooling energy by 10–15%. Pays for itself in 2.5–3.5 years.

Saves $100–$145/year + significant emissions

Low-Flow Plumbing Fixtures

Low-flow faucets, showerheads, and dual-flush toilets. Directly counteracts the water consumption AI infrastructure requires.

~$100 upfront, 2-year payback

Sustainable Material Choices

Reclaimed wood, low-VOC paints, locally sourced finishes, secondhand and vintage furniture. Reduces both emissions and indoor air pollutants.

4–8 year payback on energy savings

Rainwater Harvesting

For irrigation and non-potable uses. Reduces draw on municipal water supplies — the same ones data centers compete for.

~$3,000 upfront, ~$4,800 annual savings

Passive Design Strategies

Window placement, insulation, and HVAC optimization reduce the energy load of every building for its entire lifespan. Good design pays environmental dividends for decades.

Long-term reduction in energy demand

Jenna also notes: skip the dryer when you can, carpool, reduce food waste. These are not design-specific, but they are high-impact and immediately available to everyone listening. Hanging laundry to dry even a few times a week has more environmental impact than a year of daily AI use.

AI Can Make Your Design Practice More Sustainable, Not Less

This is the part of the sustainability conversation that rarely gets addressed: AI tools are not only an environmental cost. Used well, they reduce resource consumption and waste in the design process itself.

According to a 2023 SBE Council report and 2024 data from Marketing Profs, small businesses using AI reported saving an average of 13 hours per week and seeing a return of approximately $7,500 annually. For designers, those efficiency gains translate directly into less waste: fewer sourcing errors that lead to incorrect orders, better layout accuracy that reduces material overage, faster research on sustainable certifications and energy-efficient products, and more time for the creative judgment that leads to better design decisions overall.

"AI can help us make smarter, more eco-friendly decisions without the burnout. The carbon impact of using AI is tiny compared to the energy we save by making better design decisions faster."

— Jenna Gaidusek

The framing Jenna returns to throughout this episode: being a sustainable designer in the age of AI does not mean rejecting the tools. It means using them intentionally — for the purposes they genuinely serve — while making the other choices in your life and practice that have real environmental leverage.

Frequently Asked Questions

Using a large language model like ChatGPT ten times per day for an entire year generates approximately 16 kilograms of CO₂. For comparison: driving a gas car one mile generates about 0.44 kg of CO₂, one 10-minute hot shower generates about 2 kg, and running a load in a tumble dryer generates about 1 kg. Your entire year of daily AI use equals roughly eight hot showers, 16 dryer loads, or less than 40 miles driven. These numbers are standardized estimates based on average energy grid composition and will vary depending on where the data center is located and what energy source it uses.

Carbon emissions from AI use are real but relatively small compared to other daily activities. Water consumption is different: it is location-specific, the water is often lost permanently through evaporation rather than returned to the system, and it draws from the same municipal drinking water supplies that serve residential homes and agriculture. In water-stressed regions like Phoenix, the strain is already severe. Additionally, only 16% of major data center companies publicly disclose their water management plans, meaning most of this consumption happens without public accountability. The good news is that closed-loop liquid cooling and reclaimed water systems exist and are being adopted — but slowly and without regulatory requirement in most jurisdictions.

Switching all home or studio lighting to LED bulbs saves approximately 433 kg of CO₂ per year — 27 times the annual carbon footprint of using an LLM ten times a day. This is also one of the most universally applicable recommendations in any client project. Beyond that, smart thermostats (10–15% reduction in heating and cooling energy), low-flow plumbing fixtures, and sustainable material specifications each have significant impact. Passive design strategies — window placement, insulation, HVAC optimization — have the longest-lasting environmental payoff of anything a designer can influence.

Yes, in several concrete ways. AI tools can significantly accelerate research on sustainable materials, certifications, and energy-efficient products — reducing the time required to find and evaluate options. Faster sourcing with fewer errors means fewer incorrect orders and less material waste. Better layout accuracy reduces overage and scrap. AI-powered project management tools help catch mistakes earlier in the process, before materials are purchased or installed. And freeing up 13 hours per week of administrative time (the average reported in 2023 SBE Council research) means more time for the thoughtful creative judgment that results in better design decisions overall.

Jenna points to two specific areas: local government engagement and platform accountability. For local government: if you live in a region with water scarcity, connect with your local officials about requiring water balance studies and sustainable cooling mandates before new data centers are permitted. Screaming into the void about national policy is less effective than showing up at local planning meetings where decisions are actually made. For platform accountability: choose digital services from companies that publicly disclose their water and energy use and have credible commitments to improvement. Currently only 16% of major data center companies disclose this information — consumer and business pressure matters.

Episode 50

50 Episodes In — Thank You for Being Here

This milestone felt like the right moment to step back from the day-to-day tools conversation and talk about something bigger. The future of design requires both technological fluency and environmental responsibility. The two are not in conflict — they are the same job.

Join The DAIly Certificate Program ↗

Disclaimer: This blog was written using AI as a recap from the recording then edited by the author for accuracy and details.