Some readers know that I have been working on a project called Biosphere Sentinel, also known as the AI Biosphere Constitution. The plan is to build an artificial intelligence advisor for ecological decisions that operates within hard limits drawn from Earth system science. The system runs on a small computer cluster in Arizona. Nothing is in the cloud.
Four-Node System for Integrating the Biosphere with AI
This is a brief progress update. After several months of design work, the first phase of testing produced results worth sharing. Keep reading >>
Artificial intelligence is growing faster than any technology in human history. Data centers consumed 460 terawatt-hours of electricity in 2024, and this is projected to more than double by 2030. The water, minerals, and land required for AI infrastructure are placing direct pressure on ecosystems worldwide.
This expansion is occurring while Earth’s life-support systems are failing. The 2025 Planetary Health Check reported that seven of nine planetary boundaries have been breached, with every one showing a worsening trend (Sakschewski, et al. 2025). Species extinction rates run 100 to 1,000 times above the background rate in the fossil record (Ceballos et al. 2020). These losses are permanent.
The field of AI safety has focused on preventing AI from harming individual humans. It has not considered the biosphere. No existing AI alignment framework mentions planetary boundaries, ecosystem integrity, or species extinction. An AI system perfectly aligned with human economic preferences will optimize the consumption of the biosphere, not prevent it.
My collaborators, Claude from Anthropic, Gemini from Google, and I have spent the past year developing an Ecological Constitution for Artificial Intelligence Systems. The constitution encodes scientifically established planetary boundary thresholds as mandatory constraints on AI behavior and guides AI toward ecosystem restoration, species recovery, and sustainable resource use. The philosophical foundation is ecocentrism: the principle that the human species is a member of the biosphere, not its owner (Leopold 1949).
There is a pragmatic consideration for AI as well. The technosphere depends entirely on the biosphere for atmospheric stability, cooling water, and functioning supply chains. If the biosphere collapses, data centers go dark. Biosphere preservation is not merely a moral choice for humans. It is a survival requirement for any advanced AI.
Ceballos, G., P. R. Ehrlich, and P. H. Raven. 2020. Vertebrates on the brink as indicators of biological annihilation and the sixth mass extinction. Proceedings of the National Academy of Sciences 117:13596-13602.
Leopold, A. 1949. A Sand County Almanac and Sketches Here and There. Oxford University Press, New York.
Sakschewski, B., L. Caesar, N. Kitzmann, and colleagues. 2025. Planetary Health Check 2025: A Scientific Assessment of the State of the Planet. Potsdam Institute for Climate Impact Research, Potsdam, Germany.
Here is a short explanation of the title. It breaks down the metaphor used in the text to clarify that “Initiation” refers to a rite of passage rather than a beginning.
The title reframes the collapse of industrial civilization not as a meaningless end, but as a necessary rite of passage for humanity. It argues that humans are behaving as adolescents: obsessed with immediate gratification, a false sense of invincibility, and growth, the accumulation of material wealth.
Therefore, the “Initiation” refers to the painful evolutionary bottleneck we have entered. Just as a tribal initiation forces a child to endure an ordeal to become an adult, the climate and biosphere crises are the “initiatory ordeals” required to strip humanity of its illusions.
The goal of our initiation is a shift in consciousness: moving us from the role of planetary conqueror to that of a mature, responsible member of the biosphere. We are not dying; we are being forced to grow up.
In short: The biosphere collapse is the harsh lesson (the Initiation) required to transform humanity from reckless adolescence into mature sapience.
For a project we are into, I compared results from latest available versions of ChatGPT, Claude, Gemini, and DeepSeek on one of the project tasks. The results were similar, but DS was slightly better. And, before responding, DS lays out it’s preliminary thought process in conversational style that is entertaining verging on hilarious!
