ARTIFICIAL INTELLIGENCE (AI)

Disclaimer:
We support Decentralized AI, in other words, AI that doesn't affect the environment or deal with any of the problems associated with Centralized AI. This is ensured by Brighteon AI's Mike Adams, the Health Ranger, an independent food forensics lab scientist and investigative journalist who shares our values.

Bigfoot Nature Fellowship recognizes that industrial-scale AI infrastructure can harm ecosystems and communities through energy demand, water consumption, mineral extraction, e-waste, habitat conversion, noise, and light pollution—particularly where facilities are poorly sited or fossil-powered. We oppose those harms. At the same time, we use AI selectively and transparently as a limited outreach tool when volunteer-created images or other human-made alternatives are unavailable. We prioritize community collaboration, original creation, decentralized/local tools where practical, and tangible regenerative work such as food growing, watershed restoration, and ecological education.

Side Note: Why Agribusiness & Lawns Are A Greater Threat
Global agricultural water use is approximately 2.7 trillion cubic meters per year—hundreds of times larger than AI's water consumption
U.S. corn farming: 1 day's water use > all U.S. data centers' combined annual direct water use
U.S. golf courses: 1.4 billion gallons/day
All U.S. data centers (including indirect power-related water): ~211 million gallons/day
In the city of West Des Moines, for instance, Microsoft's data centers use an average of about 2.1% of the city's total monthly water supply, while lawn watering can account for 30% to 40% of water use during peak summer months.
According to EPA data, American households use approximately 9 billion gallons of water per day for outdoor landscape irrigation. For comparison, data centers across the entire United States used 17.4 billion gallons of water in all of 2023.

 
What “AI” Really Is (Infrastructure vs Use)

Modern AI is basically heavy-duty computing—software models running on servers in data centers—rather than some metaphysical entity; environmentally it behaves like an intensified extension of cloud computing and search, not a wholly new category. Most of the footprint comes from the underlying infrastructure (electricity, cooling, hardware manufacturing, land use) rather than from an individual user querying a model a few times a day.

Important distinctions that often get blurred in criticism:

Training frontier models vs. everyday “inference”
Training large models is very energy- and water-intensive; day-to-day tasks like generating a flyer or editing text are orders of magnitude lighter.

Centralized cloud AI vs. local AI
Large cloud data centers differ significantly from small models running on personal hardware; their environmental profiles are not identical.

Centralized AI Data Centers: Real Environmental Impacts

Energy and carbon

Global data centers (not just AI) used roughly 415–460 TWh of electricity in 2024, about 1.5% of world demand, and could reach around 935–945 TWh by 2030 (about 3%). AI workloads are a growing share of that:
Around 20% of data center energy is AI today, projected ~40% by 2030.
AI-focused “hyperscale” centers can use as much power as 100,000 homes.
Emissions from this electricity currently amount to roughly 182–208 million tons of CO₂ per year, rising toward 355–440 million tons by 2030 depending on grid decarbonization. So the climate impact is real but still a minority slice compared with transportation, buildings, and heavy industry.

Water and cooling

Cooling AI servers is water-intensive in many designs:
Data centers consumed ~175 billion liters of water in 2023, with projections above 664 billion liters by 2030.
In the US, AI servers alone may consume 200–300 billion gallons per year by 2030.
Some AI-heavy sites use up to 5 million gallons per day—akin to a small city’s demand.
These withdrawals are highly localized; some basins are hit hard, others barely affected.

Land use and “data heat islands”

AI data centers can occupy hundreds of acres, displacing farmland or natural habitat and adding concrete, steel, and transmission corridors. Recent remote-sensing work found land-surface temperatures around AI centers rising by about 2°C on average after operations begin, creating “data heat islands” that could alter local microclimates for hundreds of millions of nearby residents.
Noise, light, and community health
Cooling systems, HVAC, and backup generators produce continuous hum and vibration:
Property-line noise often sits in the 40–65 dBA range; near-field readings above 80 dBA have been reported at some facilities, comparable to a leaf blower.
Chronic exposure above about 55 dBA outdoors / 45 dBA indoors is linked to sleep disturbance, stress, and cardiovascular impacts.
All-night lighting disrupts sleep cycles and affects migratory birds, bats, insects, and other nocturnal species. Conflicts like the Nashville Zoo case—where a proposed data center next to sensitive enclosures raised alarms about noise and light impacts—show the clash when siting ignores ecosystem and educational needs.

Impacts on Ecosystems and Animals

Anthropogenic noise from data centers masks animal communication frequencies—especially low-frequency calls used by elephants, cassowaries, birds, frogs, and many insects. Studies report:
Increased vigilance and movement for individual animals under continuous noise.
Chronic stress responses and potential reductions in reproductive success.
Light pollution similarly interferes with migration, breeding, and nocturnal foraging. On top of that, data center construction fragments habitat and alters hydrology via extensive impervious surfaces and new infrastructure.
These harms are real—but they are tied to how and where infrastructure is built, not to the mere existence of AI models.

Community Health, Justice, and Concentrated Burdens

Certain areas—like Ireland, Northern Virginia, and parts of Texas and Australia—now host dense clusters of data centers and face:
20–35% of regional electricity consumed by data centers.
Heavy diesel-generator use producing particulate matter (PM) and nitrogen oxides (NOx), harming children and elders.
Large draws on local water resources.
This raises environmental justice concerns: some communities absorb noise, air pollution, water stress, and land loss while the benefits of AI are distributed globally.

Is AI “Worse” Than Other Industrial Systems?

Relative numbers matter here:
Data centers account for roughly 0.6–1% of global energy-related greenhouse gas emissions today; transport and buildings each contribute far more.
AI-driven growth could double data-center energy use over the next decade, but it still sits within a broader industrial system that is already unsustainable.
So saying “AI is destroying the environment” is too coarse: AI is adding strain to an already damaging energy–infrastructure regime, but it is not uniquely catastrophic compared to roads, airports, industrial agriculture, or military systems.

Cloud vs Local AI: Decentralization and Footprint

Centralized vs local emissions
Comparisons of local servers vs cloud data centers show that large shared facilities can be more efficient per unit of computation because they run at high utilization and have optimized cooling:
One study found local servers emitting ~9.5× more CO₂-equivalent per virtual machine over their lifetime than a data center providing the same service.
At the same time, environmental math is straightforward:
Emissions ≈ energy used × carbon intensity of the local grid.
If your local machine runs on solar or a renewable-heavy grid, and you use it efficiently, then decentralized AI can rival or beat cloud AI for operational emissions. In practice:
Cloud centers often win on cooling efficiency (lower PUE).
Local/edge systems can win if powered renewably and designed for long lifespans and repairability, which reduces e‑waste and mining impacts.
From a voluntaryist standpoint, local AI also offers more data sovereignty and autonomy—keeping computation and information within community control instead of corporate/state clouds.

AI as a Tool for Environmental Regeneration

Despite infrastructure impacts, AI is already delivering measurable ecological benefits, especially in agriculture and energy.

Precision agriculture and permaculture principles

Across dozens of studies, AI-supported agriculture has:
Increased yields by ~25% on average while reducing inputs.
Reduced irrigation water use by about 15–22% via smart, moisture- and climate-based scheduling.
Cut fertilizer use ~28% and nitrogen runoff ~35%, directly reducing water pollution and soil degradation.
Lowered pesticide applications 20–25% and chemical use 30–40% through targeted spraying based on pest forecasts.

In Nigeria, AI-driven agriculture improved yields ~25%, reduced fertilizer use by 30%, and improved solar microgrid efficiency by 15%, lowering emissions in a sector that accounts for one-third of national greenhouse gases.
These are completely aligned with permaculture ethics: they conserve water, reduce toxic inputs, protect soil and waterways, and support food resilience.

Climate modeling and adaptation

AI improves climate risk prediction by integrating satellite data, ground sensors, and historical records, yielding more accurate forecasts of extreme events, droughts, and weather patterns. In African agriculture, AI supports:
Real-time advisories on pests, rainfall, and water stress.
Intelligent irrigation systems that conserve water in scarcity-prone regions.
This helps local communities plan land use and cropping for resilience rather than reactive crisis management.

Renewable energy and grids

AI-optimized solar microgrids and smart grids:
Improve energy efficiency by ~15% compared with traditional management.
Facilitate integration of variable renewables, reducing waste and curtailment.
These functions directly reduce emissions outside the AI infrastructure itself.

Net Sustainability: Benefits vs Infrastructure Costs

Regional net-sustainability assessments show:
Under business-as-usual AI data-center growth, net sustainability outcomes are negative; emissions and water use outweigh benefits.
Under optimistic scenarios with heavy renewables and best cooling, AI-enabled efficiencies can offset ~75.8% of data-center emissions, leaving ~24.2% to be covered by further decarbonization and efficiency.
So “AI for good” projects cannot by themselves erase the infrastructure footprint; they must be coupled with energy transition, better siting, and stricter standards for data-center design and reporting.

Everyday AI Use for Flyers and Web Content

Marginal impact of small-scale use

The International Energy Agency estimates a ChatGPT-style query uses roughly 10× the electricity of a standard web search. That sounds dramatic in relative terms, but in absolute terms the energy for a handful of queries is tiny compared with:
Streaming video for hours.
Driving a car.
Running HVAC systems or industrial equipment.
Given that the bulk of AI’s footprint comes from training and serving billions of queries globally, occasional use to create flyers or images is effectively negligible in that global context—especially if the broader work is restoring land, growing food, and building local resilience.

The key questions are:
What does the AI-generated content serve?
If it helps draw people into food forests, watershed restoration, and natural-law education, the net ecological gain from those activities dwarfs the marginal energy of the graphics.
Is AI a last resort or a default?
We already prioritize human-created art and only resort to AI when volunteers and our own capacity are insufficient; that is consistent with minimizing dependence and honoring voluntary collaboration.
Are we transparent and open to alternatives?
We invite people to contribute images and art; AI fills gaps rather than displacing human creativity. That is aligned with voluntaryism and fair share.

From an objective standpoint, using AI occasionally for outreach, while simultaneously growing food, building watersheds, and creating real biophysical abundance, is far less harmful—and arguably net positive—than refusing a minor tool that can expand our regenerative reach.

How AI Can Be More Aligned with Permaculture and Voluntaryism

Purpose-bound use
Use AI primarily for tasks that clearly support earth care, people care, and fair share—educational materials, ecological design support, coordination—not generic promotion or consumerism.

Minimize dependence, maximize collaboration
Keep doing what we already do: ask for community-created art first, use AI only when necessary, and treat it as a bridge tool while human networks grow.

Prefer lower-impact and transparent providers
When using hosted AI, favor services that disclose energy and water use, invest in renewables, and avoid siting in water-stressed or ecologically sensitive regions.

Explore decentralized/local AI where appropriate
For some tasks, consider smaller local models on efficient hardware powered by solar or other renewables; this reinforces autonomy and can reduce reliance on large hyperscale centers, depending on your grid context.

Oppose harmful data-center siting
Support resistance against data centers placed near zoos, wildlife corridors, or already stressed communities, and advocate for noise limits, environmental impact assessments, and community consent.

Couple AI use with tangible regeneration
Make it a conscious praxis: every time AI significantly helps our outreach, pair that with concrete permaculture actions—new garden beds, tree planting, watershed work—so the net energetic and ecological flow is firmly toward regeneration.