The century of catastrophe and change: the climate crisis and artificial intelligence

Many centuries have been centuries of catastrophe, and every century has been a century of change. In the 20th century, 187 million people were killed in conflict, it was a century of intense catastrophe; the Industrial Revolution, globalisation, decolonisation and changing geopolitical hierarchies made it a century of intense change.

But the 21st century is different.

It is plausible that things as simple as dirty air and intense heat will kill as many people as conflict did in the previous century, the most murderous on record. And AI will transform the world in ways the Industrial Revolution couldn’t imagine.

The 21st century will almost certainly be a century of unmatched catastrophe and unmatched change.

We are in the back end of a closing window of opportunity to prevent the worst catastrophes and to make sure the change that happens is in humanity’s interest. The actions we do and do not take will be decisive.

The main drivers of catastrophe and change? The climate crisis and artificial intelligence (AI).

The climate crisis is made up of sub-challenges that are often mutually reinforcing, fiendishly complex, lethal, and together they are existential. AI is developing at unprecedented speed and scale, it is transformative, it feels uncontrollable, and it is proving intensely disruptive, for good and for bad. Both are pushing our systems and structures, and the people who bring them to life, to their limits.

The climate crisis and catastrophe

The UN estimates we are on track for 2.6 to 2.7 degrees warming by the end of the century, and 3.2 degrees warming if polices are not strengthened.

The decade from 2010 to 2020 was 1.07 degrees warmer than 1850-1900; global average temperature is increasing and catastrophe is already hitting every corner of the world. Take these three BBC news stories from the past 7 days:

And it is going to get worse before it gets better.

If global temperature increases by just two degrees above pre-industrial temperatures, 150 million more people would die from air pollution alone and the world would be $20 trillion poorer compared to a 1.5-degree warmer world. 400 million people would suffer from water scarcity. There would be 32 times as many extreme heat waves in India, and each would last five times as long, exposing 93 times more people. Sea level will rise by about two to three meters if warming is limited to 1.5 degrees and 2–6 meters if limited to two degrees.

It is a catastrophic best-case scenario.

If economic growth is better than anticipated or if climate action stalls, three degrees warming by 2100 is likely. In this scenario, the average drought in North Africa would last five years longer than the average today. Southern Europe would be in permanent drought. Wildfires would burn twice as much land in the Mediterranean and six times more land in the United States. Because of the compound interest of climate catastrophe, 3.7 degrees warming could produce $550 trillion in damages, around 20% more than total global wealth today.

The World Bank estimates there will be 140 million climate refugees by 2050, more than 100 times Europe’s migration crisis of Syrian refugees. The UN’s estimate is closer to 200 million refugees. A billion or more vulnerable poor people could be left ”with little choice but to fight or flee”.

Imagine one in nine people on Earth waking up every morning having to decide whether to fight to survive in conditions that humans have not evolved to live in; or to flee their homes, travelling in these inhospitable conditions just to arrive at a country that probably demonises you; or to risk your life in conflict just for a drink of water.

Much of the important stuff in the crops we grow – protein, calcium, iron, vitamin c – has declined by as much as one-third since 1950. Nutrient collapse means that by 2050, up to 150 million people in the developing world will be at risk of protein deficiency, since so many of the world's poor depend on crops, rather than animal meat for protein. 140 million could suffer from a deficiency of zinc, essential for healthy pregnancies.

Even if climate action radically increases, the best-case scenario still presents a scale of catastrophe that is new to us humans and it will push us and everything we have built to the limit.

The battle to live in conditions that lie outside the Goldilocks temperature range that made life possible, and which no intelligent life has survived outside before, will define this century.

Mitigation and adaptation need to be the centripetal force around which our public and private sectors orient themselves if we are to have a chance of fighting off the worst of the catastrophes that could come.

AI and change

In December 2022, ChatGPT was launched. AI became the big thing. ChatGPT gave people a way of feeling the power of AI in a much more tangible way than AI’s more obscure and hidden applications that are very powerful, but a few steps removed from user interfaces and public consciousness.

The speed at which AI systems are developing, the potential power they possess, and the sheer volume of applications AI has make it a technology that will make this century one of unmatched change.

OpenAI launched GPT-1 in 2018. It had 117 million parameters, a measure of the systems scale and complexity. GPT-4, launched in March of this year, has over a trillion parameters. Models exceeding 100 trillion parameters, the number of synapses in the brain, could be with us in the next five years.

AI will soon be changed by AI’s own self-development capabilities, it will improve itself and learn how to do new things, without human intervention.

It will change the world.

There are three key sites of change: geopolitics, governance and work.

Geopolitics

Creators of AI and associated advanced technologies are becoming geopolitical actors. Technology companies ”wield a kind of power in their domains once reserved for nation states” write Ian Bremmer and Mustafa Suleyman in their recent piece on AI governance.

Elon Musk decided not to extend Starlink coverage to Crimea at the emergency request of the Ukrainian government which prevented a Ukrainian attack on Russian forces. An unelected tech billionaire is making decisions that directly affect the trajectory of a war between two nation states.

As AI and advanced technologies become further embedded into critical infrastructure, those who control these systems become immensely powerful absent proper oversight.

Bremmer and Suleyman argue that to govern AI, a global framework is needed that will require the international system to move beyond ”traditional conceptions of sovereignty and welcome technology companies to the table”. Technology is creating new geopolitical actors.

Whichever state achieves an artificial general intelligence (AGI) breakthrough could win itself an era of geopolitical dominance ”not unlike the… nuclear superiority the United States enjoyed in the late 1940s” (Eric Schmidt).

AGI would be able to perform any mental task a human can, at a minimum. It would bring dominance in scientific and technological domains. AGI is not imminent, but it could very plausibly be created well within the next fifty years. The stakes are high for the US and China, the two most likely candidates to get this breakthrough.

Governance

AI will change feedback systems that are critical to any type of national governance system. It has different effects on democracies and autocracies.

In democratic systems, negative effects of AI include increasing polarisation when instructed to create addictive social media feeds that have the effect of affirming existing beliefs and creating deepfakes that are indistinguishable from reality undermining people’s ability to make political decisions from trustworthy information.

Positive effects include improving the quality of information politicians and civil services use to make decisions in line with the electorate and improving understanding of how public policy can be used to solve inherently complicated problems with public policy.

In autocratic systems, AI might seem like a tool for creating more detailed and accurate representations of political beliefs than democratic elections do. But the biases embedded into algorithms could have the effect of concealing disagreement from decision makers and reinforcing their own beliefs about the world to the point where they find themselves in an ”AI-fuelled spiral of delusion” as Farrell, Newman and Wallace write.

It is unclear which of AI’s positive and negative effects on governance systems will become the most important effects, and it is unclear what it will mean for systems as a whole once all the positive and negative impacts are added up. What is certain is that AI will change fundamental characteristics of governance systems.

Jobs

Research from OpenAI and others estimates that nearly 50% of all US jobs could have half their functions performed by AI; McKinsey found 50% of global work activities are technically automatable with current technologies. Professor Diane Coyle argues that economists mostly agree that historically, the companies that have utilised automation increased job numbers and paid higher wages. Automation had the effect of changing where humans were needed from blue collar jobs to white collar jobs, it did not replace humans.

What makes AI different is that it is taking the white-collar jobs as well as increasing the range of blue-collar jobs that can be automated, or made easier which reduces wages. The thing that protected human labour was that we could always think better than machines. That will soon no longer be the case.

But AI and advanced technologies also create jobs. Labour markets tend to adapt to new technologies over time preventing mass unemployment. The key impact of AI on work is likely to be that workers will need to transition into different kinds of jobs, rather than transitioning out of jobs, but this will vary heavily across sectors. The overall effect could be on the scale of the shift out of agricultural jobs in the early 1900s.

McKinsey estimate that between 400 million and 800 million people could be ”displaced by automation and need to find new jobs by 2030”, and that 75 million to 375 million people may need to ”switch occupational categories and learn new skills”. This will be destabilising and could have significant political implications if governments do not heavily invest in protecting workers’ rights and re-skilling to smooth the transition into different types of work.

AI will change geopolitics, governance and work. If AI is not governed effectively, the century or catastrophe and change could become the century of catastrophe and catastrophe. Once a claim of the most sceptical, now a line in Ursula von der Leyen’s State of the Union speech on Wednesday 13th September 2023: ”mitigating the risk of extinction from AI should be a global priority alongside other societal risks such as pandemics and nuclear war”.

What does it mean to live in the century of catastrophe and change?

Hundreds of millions will die early, quality of life for billions will not be good, some things will radically improve, some things will radically change with neutral effect, some things will radically worsen, not much will stay the same.

There is a chance that if, by 2100, we succeed in transforming the bedrock of our world from fossil fuels to renewable energy, planting millions of trees, reconfiguring our transportation systems and food systems, reducing our consumption and governing AI effectively so we reap its benefits while avoiding its most negative potential impacts, the world might be in a place to make 2100 the century of health and prosperity.

But to make it to 2100 in reasonable enough shape for that scenario to be even a remote possibility, we first need to deal with this century of catastrophe and change.

That is as demanding a task as humanity has ever faced.