When we were discussing how AI could be demonstrated, one issue was that people have difficulty to grasp what intelligence is and how to measure it. Everyone knows that intelligence is real, but the concepts associated with it cover a vast area. There is a wide gulf between machine learning, i.e. adaptation, on the one hand and recognizing fundamental concepts, for example how electromagnetic fields propagate, on the other hand.
15 years ago I worked on a novel approach to model climate change. The core idea was that the adaptation of weather patterns and its feedback mechanisms isn't solely dependent on geological and physical aspects, but rather that the global state of the Earth's ecosystems has a major influence. That there are significant feedback mechanisms between the weather and the ecosystems and that ecosystems have a built-in intelligence that adapts to the physical conditions as needed, for example stress factors induced by climate change. This is a counter to what can be described as chaos or entropy, in order to maximize an ecosystem's chances of survival. Humans unwittingly mess with these mechanisms, which is a major driver of climate change.
With other words the current climate model is too simplistic.
Describing the inherent feedback mechanisms that are a part of the ecosystems provides also insight into how intelligence works. After an introduction to the cornerstones of the model follows a short introduction of the challenges involved in modelling an AI that mirrors the behaviour of ecosystems.
In relation to climate change – and weather patterns in general – the first question we have to ask is whether ecosystems affect the weather. The second question is how does it do it.
Ecosystems undoubtly have an influence on how much water is retained after rain in the soil and how much sunlight is reflected. These factors alone have a major impact on how much groundwater is collected, how rivers are fed, and as an implication how water is distributed on land. Together with the flora's capability to reflect and absorb sunlight it also has a major impact on the temperature on the surface and how much water evaporates. Temperature is a key which is a major driver in the formation of high- and low-pressure areas and cloud formation. There can't be any doubt that ecosystems affect the weather and large scale weather patterns.
The second question, how ecosystems affect the weather patterns, is much more involved because it isn't sufficient to count all the direct factors that have an impact. We also have to ask how ecosystems adapt to changing conditions and whether this adaptation has a major impact as well, for example in the formation of cyclical weather patterns.
Here we are at the heart of the question, do ecosystems have an intelligence that they use to alter the feedback mechanisms with the weather in their favour? If so, the most likely candidate where this intelligence is located is in the genetic pool of an entire ecosystem. Something that has been honed for millions of years.
But let us first have a look at what historic evidence is available to support this theory.
Historically human civilizations began to thrive after they acquired the ability to do agriculture. The Euphrates region and the Nile delta were the first regions where this happened, followed by the Yangtze delta in China and regions inhabitated by the Maya in central America. Is this coincidence or a sign that humans worldwide acquired the intelligence to do so worldwide? Most likely not, it is much more likely that the weather patterns globally had settled into an equilibrium. As the weather patterns continued to settle more regions followed: Ireland, Crete, Greece, India, central Europe. In this context the Bronze Age collapse is most likely a temporary reversion of the weather patterns, a time of increased volatility and crop failures that in turn caused waves of mass exodus. Again, that the region least affected was the Nile delta, where agriculture happened first, is most likely not a coincidence, but shows that some regions have a better predisposition to settle into stable weather patterns.
Fifty years ago low-pressure areas travelled well-defined paths across the Atlantic. These patterns have gradually dissolved, with the polar vortex becoming more dominant. We see similar effects on the monsoon cycle on the Indian subcontinent. In terms of a geologic time scale these changes to global weather patterns happen extremely rapidly. It is questionable whether these rapid changes can be explained by changes of the CO2 level in the order of a fraction of a percent or by changes of the median temperature by a fraction of a degree Celsius.
If we look for an explanation, we have at least to consider that the cause and effect are reversed in comparison to the traditional hypothesis that the geophysical environment is the primary cause. Instead we postulate the hypothesis that the state of the Earth's ecosystems has a significant impact on the global weather patterns. In turn this means that human activities like urbanisation, deforestation and large scale pollution are the root cause of globally changing weather patterns, and that CO2 emissions are just one aspect of many. Moreover there is good reason to believe that Earth's ecosystems as they existed before massive intervention by humans, when Europe was covered almost entirely by forests, would have offset even todays CO2 emissions.
In 1985 it was estimated that a third of the world's ecosystems had either been destroyed or significantly impaired. Today it is estimated that this has happened to two thirds of all ecosystems.
If the Earth's ecosystems indeed have a major role in stabilizing the global weather patterns, then the threat we face is not so much global warming but geoentropy, an increasingly volatile and hostile environment, driven by our relentless destruction of nature.
Counter to geoentropy
There is ample scientific evidence that indicates that the ecosystems and weather patterns are intricately intertwined, from Saharan dust storms that fertilize the Amazonas and Caribbean to the level of plankton in the sea that affects albedo and evaporation. Ecosystems have a major impact on how much water an environment can retain, and affect the water cycle and temperature regulation.
It is rarely a direct link that connects the species of an ecosystem with the climate. It is not that the plankton in the sea affects its albedo directly. In the Carribean substances that are produced by planktonic algae oxidize in the atmosphere and form a major source of cloud-condensation nuclei, which in turn affects the cloud density and albedo. These clouds fulfill an important role to bring water to the Amazonas rainforest. The potential for feedback mechanisms between the ecosystems and weather patterns is immense.
If we come back to our original question, it isn't enough to simply identify feedback mechanisms. Instead the question is, whether ecosystems can indeed shape weather patterns in their favour.
If we look at the above example, the trait that planktonic algae produce a substance that evaporates and functions as cloud-condensation nuclei, would need to have evolved genetically to fulfil this role. Is this reasonable?
We need more scientific evidence to establish this. In principle these species of planktonic algae would somehow need to have better chances of survival with this trait.
If we look at individual species only, it hardly seems feasible. But if we consider entire ecosystems, including insects and bacteriae, its genetic pool is gigantic with nearly an infinite amount of potential feedback mechanisms. The biodiversity of ecosystems and its composition can significantly change during different phases, for example growth to cover a new area versus maturing. Different natural selection in response to stress is also well known. The high number of diverse insect species on Earth is another indicator how far the adaptability of ecosystems has evolved.
It is worth noting that one of the oldest cultures on Earth, the Aborigines in Australia have inherent knowledge of how certain signs in the flora or fauna are linked to phases of larger global weather patterns that can span decades. In the light of what has been discussed in the preceeding paragraphs, these skills naturally identify the adaptation of an ecosystem, and speak volumes about the Aborigines' insight.
If we consider the adaptability of ecosystems and the potential for feedback mechanisms with the climate, we can postulate the theory that the ecosystems on Earth as a whole have genetically evolved to offset geoentropy and to settle into an equilibrium. There are many reasons why ecosystems benefit from stable weather patterns. Vice versa, if it weren't for the biosphere we would see much more extreme and volatile weather.
The levels of energy efficiency and engineering feats that have evolved naturally are way beyond of what human engineers have achieved so far. This is also true of the capability of the ecosystems. Any farmer knows how difficult it can be to grow a single crop and how many setbacks can occur. On the other hand a rainforest will grow very fast to cover suitable areas. Its biodiversity is its strength, it contains all the species it needs to adapt to different conditions and phases of growth. All species fulfil a function, and taking away even a single one impacts the ecosystem's potential.
The biodiversity and its genetic pool gives ecosystems the potential to adapt to different temperatures. The role genetics play in evolving and adapting ecosystems to maintain an equilibrium is not understood today. However, there are strong indicators that there is much more to it than we know, for example the forests in the area around Chernobyl have adapted in surprisingly new ways to the radioactivity, and plants have been found to adapt much faster to parasites or a lack thereof than random genetic selection could explain.
Connecting the dots
To put all these facts into perspective, we also need to understand that the key to overcome entropy in a system is efficiency. With other words, no technology humankind ever has developed is anywhere close to restore an equilibrium to our ecosystems and global weather. This is why maintaining and restoring the natural ecosystems where we can is our best chance to avoid catastrophic levels of geoentropy. Ideally we should strive to replant forests and have the tropical forests reclaim their ground to a level seen last 1850.
Traditional theory that sees in geophysical phenomena the root cause of climate change projects only a gradual increase of volatility from the impact of global warming. If indeed the impairment of the natural ecosystems is the root cause of geoentropy, we face instead an exponentially growing threat, because the impairment of the biosphere through human expansion in the last decades has gathered pace significantly, and there is nothing to offset it.
This means the impact will hit much faster and harder than currently expected.
It doesn't take that much to wipe us out, if our fields are swamped or parched three months each year, storms ravage the harvest further, only a quarter of the pollinators survive, and fungi and viruses multiply. Once this happens our agro-industry becomes prohibitively expensive, our wealth will evaporate and we won't be able to fund R&D on the scale necessary. We have to invest into science and research now while we still can afford it. If we don't divert at least half of our military budgets to research that matters now, we may squander our chance of survival, which is probably much more slim than we believe today.
Intelligence in ecosystems
The first thing that we have to be aware of is that intelligence is always manifested in something. It isn't an abstract function.
We can approach this subject in many different ways. We can focus on behaviour and traits and link it to genetics.
Most ecosystems have food chains, i.e. a highly complex way to shape the proportions of different species and their quantity. These proportions adapt to different phases of development of an ecosystem, to stress factors, like scorching heat, and many more parameters.
All in all we have so many interdependent variables that it is a mind-boggling exercise to even imagine the capability of an ecosystem with its entire genetic pool. It's like the most complex differential equation ever.
Each trait a species has matters, whether it is how a creature reacts to a more rapid temperature fluctuation, or how the availability of nutrients affects growth. It is also necessary to distinguish between different ecosystems and how these interact with each other. This could be done geographically (their location) or functional (groups of species). The functional approach is more flexible because it allows to model ecosystems as being made up of components (different groups of species) which has advantages in modelling the interactions between different ecosystems.
In principle it is not difficult to see that an ecosystem has the capability to create a positive feedback mechanism with its local microclimate and to adapt accordingly. What is hard to see is where the intelligence is located to drive the adaptation of one ecosystem that can affect others that are thousands of kilometres away. There need to be feedback mechanisms for this to happen. But it is not unthinkable, for example ecosystems could have learned to adapt to changes in the microbiomes that could travel on currents of air or water.
The AI I am developing here is capable of modelling such complex systems. We will have a look at how this works to predict the interactions of an ecosystem with its geophysical environment in another blog post.
The idea is to use it as a tool to gather scientific evidence whether ecosystems can adapt to change global weather patterns in their favour.