Guest essay by Eric Worrall
The Conversation asks why ordinary people are so hostile towards Extinction Rebellion, when the world is on the brink of a sixth major extinction?
In the face of chaos, why are we so nonchalant about climate change?
October 19, 2021 9.37pm AEDT
Research Fellow in Politics and International Studies, University of Warwick
In the UK, for example, peaceful protest by environmentalist groups like Extinction Rebellion tends to be opposed more than it’s supported. This is despite the limited disruption these groups cause in comparison to the extreme disruption already produced and threatened by climate breakdown, such as extreme droughts, wildfires and tropical storms.
Recent protests blocking British motorways to call for the government to insulate homes have been met not with policy reform but with outrage and proposals to increase police power to arrest protesters.
So why do so many people oppose the call for change in the face of a sixth mass extinction? Why is there resignation, rather than resistance?
And I think that the lack of widespread mobilisation is borne, not from outright climate denial, but rather from a more insidious climate apathy: what might be called “climate nonchalance”.
This nonchalance – recognising the impending collapse of our world and shrugging our shoulders – is made possible only by a profound separation between the comfortable lifestyles of the privileged and the consequences of those lifestyles elsewhere: including increased death rates, frequent exploitation and environmental displacement for the less privileged.
The author appears to suggest people are too comfortable to embrace change. We do not support Extinction Rebellion because we are selfish and lazy.
But I think the answer is far simpler – human belief is a continuum.
How can the answer to a true or false question, like “is climate change a problem”, be a continuum?
As a software developer, I see this odd continuum behaviour manifest all the time, when working with artificial neural networks.
Neural networks, attempts to create an artificial intelligence which mimics the architecture of the human brain, are not places where the absolute rules. If you say attempt to train a neural network to add two numbers, it is very difficult to get an exact result. Ask a trained neural net the answer to 2 + 2, and you will receive an answer like 4.1, or 3.9, or 3.5 – anything but 4, most of the time, unless the neural net is very rigorously trained.
Similarly if you ask a trained neural network if something is true or false, you are more likely get an answer like 70% true, or 48% true. An answer of 100% or 99% true is very unusual.
Computer scientists usually deal with this kind of ambiguity from artificial neural networks by interpreting the answer. So for example, they might apply a rule that if the answer is 70% or more true, report the answer as completely true.
Obviously humans are capable of concise mathematics, so our brains are not exactly the same as artificial neural networks, but in my opinion this neural net continuum of belief manifests throughout human behaviour when you look for it.
For example, many people when asked agree that climate change is a problem. But if you ask them if they are wiling to spend even one dollar more to fix climate change, agreement plummets.
Based on what I have personally experienced when working with artificial intelligence, I believe this strange belief yet not belief is a manifestation of the human brain’s neural net continuum of belief. People might answer they believe in climate change, they believe enough to say yes, but deep down they do not believe enough to commit actual effort to solving the issue they verbally agree is a problem.
Society’s current level of almost belief is precarious – a neural net which returns an answer of 70% true can easily be trained to raise that result to 98% or whatever. Getting to 70% is far more difficult than raising 70% to 98%. In my opinion there is a real ongoing risk that people who are mildly concerned about climate change could be rapidly tipped over into fanaticism.
But training an artificial neural network to such a fever pitch of compliance requires utter silencing of all discord in the training data. Even a few discordant training samples, a handful of voices raised in disagreement, is enough to introduce doubt, to nudge the neural network away from perfect compliance.
If you achieve perfect compliance, the end result of such rigorous training is surprisingly dysfunctional. Overtraining or overfitting as AI scientists describe it, creates an artificial neural network which is far less able to cope with ambiguity or new data, than a neural network which was less rigorously trained, or was trained using noisier, more discordant data. An overtrained neural network responds perfectly to its training stimuli, but does not respond well when presented with new data (see the diagram at the top of the page).
The parallel with the human condition seems obvious.
via Watts Up With That?
October 19, 2021 at 08:42PM