Notes on reading Novacene by James Lovelock 2019

Definition of life: entities that reduce entropy, as they organise their environment

I knew his 1970s Gaia Theory, but remembered it mostly as a type of systems thinking and seeing earth as a complex system. But he adds something key:

In earth’s case the purpose of the system is to keep earth cool, to keep temperatures at 15C average. And do so as our sun slowly heats up.

A startling assumption to me is that earth really is not in the Goldilocks zone, but Mars is. We would be like Venus, hot, if not for the entropy reducing earth life. That life continuously draws down heat.
Furthermore, to an alien observer earth would not look cool but much hotter because of dumpng solar heat continuously.

The sun is heating up and so is earth therefore. Keeping cool is our prime directive. The climate urgency is making it worse and burning fossil fuells (stored heat from the past) should stop.

The Anthropocene started with the steam engine, when humans could influence their environment on a global scale. The Novacene is the coming age of AI.
The optimal temperature range for electronics and life are similar, and life and AI have the same hard upper temperature limit of 47C.
Above it we will have a runaway process to becoming like Venus.

AI will not deliberately kill us because it needs the world to stay cool under a heating sun. Carbon based life is needed for it. They will supplant us by evolution, slow not sudden, as evolution moves beyond us, as it always would.

Interesting notion: AI might become 1M times faster than us, but they are bound by the same physics as us. It means e.g. their travel will be at roughly the same speed.
Which will be 1M as boring and slower to AI than to us.

Makes a good caveat: AI would need to start its evolution from ‘good’ beginnings. E.g. not from autonomous weapons platforms.
Yet precisely in civic tech such as aviation we put hard constraints on AI. But we do not on military AI, making it more likely it will evolve from there.

My takeaway from this is how to use AI for civic tech, and set it free as it were, with a sense of communal values. Including with a sense of the Prime Directive to keep cool.

That I think is a core flaw in Lovelocks reasoning. Yes, the PD is to keep cool. Not only for our self-created heating, but mostly for the sun heating.
But how many humans are aware of this, and of those how many care enough to act, given the timescale of the suns heating in millions of years?
How will we make AI aware, and will they care where we do not, given that their relative timescale is even up to a million times longer?

He stresses the notion of the engineer and artisanal engineering. Where knowing how to make things work is a priori more important than knowing why it works.
This also ties into his notion that intuiting is key for engineering, and the scientific method of standing on the shoulders of others is more suited for the ‘know why’

Some of my takeaways:

  • When increasing the abundance of life is good to keep cool, greening your urban living environment makes sense on a deeper level than just cooling the city.
    Also as cities are an efficient way to house us humans at our current numbers.
  • How to use ML for civic tech, for networked agency
  • How to explore ML, what it currently does, what it can do, areas of issues it could be used in.
  • What autonomous things would be valuable in the home, neighbourhood, city.
  • What would an “AI in the wall” be like?