Main points from a talk I had with ChatGPT about data centers and human/environmental cost of data centers & AI. Including death.
All the bold numbered text are scaled down versions of my questions, the rest is ChatGPT's answers. I asked it to put our chat into a broken down, "shareable" version and this is it. Chatgpt also titled the whole thing.
TEN
______
We need to ask an important question:
Are we properly accounting for the environmental and human costs of the AI revolution?
The United States remains a major centre of AI computing infrastructure. The UK is also becoming increasingly important, with facilities and projects associated with London, Cardiff, North Tyneside and Essex.
These facilities are industrial operations, often running continuously, 24 hours a day.
Water can be consumed through:
The bigger concern is the cumulative consumption across an expanding global industry.
Alternatives already exist:
Their principal noise sources include:
Potential consequences include disrupted sleep, stress, annoyance and reduced quality of life.
Residents should have access to independent noise assessments, overnight monitoring, enforceable limits and meaningful complaints procedures.
Electricity generation can produce particulate pollution and ground-level ozone, contributing to respiratory and cardiovascular illness.
A November 2026 MIT research estimate attributed approximately 650 premature deaths in the United States in 2023 to air pollution associated with additional electricity generation for data centres.
This is a modelled estimate, not a count of individually identified deaths.
Britain needs equivalent research to understand the potential health burden here.
These deaths deserve recognition and proper investigation.
These cases raise serious questions about AI safety, responsibility and safeguards.
However, an allegation of AI involvement is not the same as proof that AI caused a death. Careful investigation is essential.
We should be asking governments and technology companies to:
But economic growth and technological progress should not come at the expense of clean water, clean air, peaceful homes, workers' safety or human life.
We measure the success of AI in computing power, investment and financial returns.
We should also measure its success by its impact on people and the environment.
The question is not whether we should have artificial intelligence. It is whether we can develop it responsibly, transparently and without imposing unacceptable costs on society.
An informed public is essential to responsible technological progress.
All the bold numbered text are scaled down versions of my questions, the rest is ChatGPT's answers. I asked it to put our chat into a broken down, "shareable" version and this is it. Chatgpt also titled the whole thing.
TEN
______
THE HIDDEN COST OF ARTIFICIAL INTELLIGENCE
Data centres, water, noise, pollution and human life
An examination of the environmental and human consequences of the AI revolution
Artificial intelligence is transforming our world, but behind every AI conversation lies an enormous physical infrastructure of computers, electricity, cooling systems and industrial facilities.We need to ask an important question:
Are we properly accounting for the environmental and human costs of the AI revolution?
1. Where does the infrastructure exist?
AI services such as ChatGPT depend on enormous data centres containing thousands of powerful computer processors.The United States remains a major centre of AI computing infrastructure. The UK is also becoming increasingly important, with facilities and projects associated with London, Cardiff, North Tyneside and Essex.
These facilities are industrial operations, often running continuously, 24 hours a day.
2. Water consumption
Data centres generate enormous quantities of heat and need sophisticated cooling systems.Water can be consumed through:
- Evaporative cooling systems.
- Industrial cooling towers.
- Electricity generation supplying the facilities.
- Manufacturing the computer equipment.
The bigger concern is the cumulative consumption across an expanding global industry.
Alternatives already exist:
- Closed-loop liquid cooling.
- Dry-air cooling.
- Seawater heat exchange.
- Recycled wastewater.
- Rainwater harvesting.
- Immersion cooling.
3. Noise pollution near residential areas
Data centres operate around the clock.Their principal noise sources include:
- Industrial cooling fans.
- Chillers and ventilation systems.
- Electrical transformers and substations.
- Emergency diesel generators.
Potential consequences include disrupted sleep, stress, annoyance and reduced quality of life.
Residents should have access to independent noise assessments, overnight monitoring, enforceable limits and meaningful complaints procedures.
4. Air pollution and premature deaths
The environmental impact extends beyond the boundaries of individual data centres.Electricity generation can produce particulate pollution and ground-level ozone, contributing to respiratory and cardiovascular illness.
A November 2026 MIT research estimate attributed approximately 650 premature deaths in the United States in 2023 to air pollution associated with additional electricity generation for data centres.
This is a modelled estimate, not a count of individually identified deaths.
Britain needs equivalent research to understand the potential health burden here.
5. Construction accidents and human life
Building large data centres involves:- Heavy construction equipment.
- Electrical installations.
- High-voltage substations.
- Specialist contractors.
- Extensive industrial infrastructure.
These deaths deserve recognition and proper investigation.
6. Deaths associated with AI systems
There have also been reported cases in which families have alleged that interactions with AI chatbots contributed to suicide or other fatal incidents.These cases raise serious questions about AI safety, responsibility and safeguards.
However, an allegation of AI involvement is not the same as proof that AI caused a death. Careful investigation is essential.
7. What should we demand?
The expansion of AI should come with meaningful environmental and public-health protections.We should be asking governments and technology companies to:
- Publish annual water consumption figures for individual data centres.
- Disclose electricity consumption and associated carbon emissions.
- Prioritise recycled water and closed-loop cooling.
- Conduct independent noise assessments, including low-frequency noise.
- Protect nearby residential communities from continuous industrial noise.
- Publish workplace accident and fatality statistics.
- Investigate the health consequences of pollution from AI-related electricity generation.
- Provide transparent reporting on AI-related safety incidents.
- Ensure local communities have a meaningful voice in planning decisions.
CONCLUSION: TECHNOLOGY MUST SERVE HUMANITY
Artificial intelligence has enormous potential to benefit society.But economic growth and technological progress should not come at the expense of clean water, clean air, peaceful homes, workers' safety or human life.
We measure the success of AI in computing power, investment and financial returns.
We should also measure its success by its impact on people and the environment.
The question is not whether we should have artificial intelligence. It is whether we can develop it responsibly, transparently and without imposing unacceptable costs on society.
An informed public is essential to responsible technological progress.