Feature Request: Make the Environmental Impact of Every AI Request Visible
Every AI request has a physical footprint.
It consumes electricity, requires computing hardware and cooling, and can involve water consumption depending on the data centre and energy infrastructure. Yet from the user’s perspective, AI appears almost completely weightless.
I propose that ChatGPT introduce a standard AI Environmental Impact Indicator showing users an estimated environmental footprint for their requests.
For example:
Energy: ~0.8 Wh
Water: ~5–15 mL
Carbon: ~0.2–0.5 g CO₂e
Impact: Low
Confidence: Medium
The numbers should be estimates and ranges, not false-precision measurements. OpenAI should publish the methodology and assumptions behind them.
The feature could include:
1. Environmental impact before or after a request
Users could see the estimated energy, water and carbon associated with processing their request.
2. A simple traffic-light indicator
Low
Moderate
High
Very High
3. An “Eco Mode”
Where appropriate, ChatGPT could favour more computationally efficient approaches, shorter outputs or smaller models.
4. Suggestions for reducing impact
For example:
“This is a large request. You could reduce computational demand by using a shorter response or removing unnecessary attachments.”
The user would always remain in control.
5. A cumulative personal footprint
For example:
Your AI use this month
42 Wh
~0.6 L water
~18 g CO₂e
17 requests
6. Organisational reporting
Universities, businesses and other organisations could eventually see aggregate AI energy, water and carbon estimates.
Most importantly: transparency
OpenAI should not claim an exact environmental cost when it cannot measure one precisely.
The system should show uncertainty and explain the assumptions used.
This could eventually lead to an open AI Environmental Impact Standard, developed with researchers, universities, data-centre operators and other AI companies.
The objective isn’t to tell people not to use AI.
It is to make the physical consequences of AI visible so people can make informed decisions about how they use it.
We routinely receive information about the financial cost, time and output of digital services. We should also have meaningful information about their environmental cost.
Every AI request has a physical footprint. Users should be able to see it.
I encourage OpenAI to investigate making environmental-impact information a standard component of ChatGPT and to work toward an open methodology that could become an industry-wide standard.