The Republican Party is meek, and Trump knows it. He would not have dared to send Senate Republicans names like Robert F. Kennedy jnr, Tulsi Gabbard, Kash Patel and Pete Hegseth for Cabinet posts in his first term. Even beyond the party, he faces no mass resistance this time, nothing like the Women’s March that overwhelmed Washington in 2017. Democrats are dispirited and exhausted.

Trump is now flanked by an alliance of oligarchs led by Elon Musk. The billionaire owners of the Washington Post and the Los Angeles Times killed presidential endorsements of Kamala Harris, ABC News (owned by Disney) gave Trump’s “future presidential foundation and museum” US$15 million ($26.9m) to settle a defamation lawsuit Trump brought, Mark Zuckerberg is refocusing Meta platforms around “free expression” and his company against diversity, equity and inclusion, and Amazon reportedly paid US$40m for Melania Trump’s documentary about herself. Tim Cook, Sundar Pichai and a slew of other cheif executives have recently travelled to Mar-a-Lago in Florida to dine with Trump. This differs from 2017, when Trump was treated as an aberration to be endured or a malignancy to reject. The billionaires see the rules have changed. They are signalling their willingness to abide by them. “EVERYBODY WANTS TO BE MY FRIEND!!!” Trump wrote on Truth Social after having dinner with Jeff Bezos. He isn’t wrong.

Democracy does not die in darkness. It degrades through deal-making – a procession of pragmatic transactions between those who have power and those who want it or fear it. We have seen this termitic corruption consume democracies elsewhere, including Viktor Orban’s Hungary, which Trump and his allies cite as a model. Money and media make peace with the regime because to do otherwise is too costly. Once-proud political parties become vehicles for the individual ambition they were designed to replace. I consider the range of outcomes for Trump’s second term to be stupefyingly vast, stretching from self-destructive incompetence to muddling incoherence to authoritarian consolidation. But the levees that narrowed the possibilities of his first term have been breached.

Startling gains in AI

As I was writing those paragraphs, I asked ChatGPT three or four questions. They were small queries of the sort I might once have asked Google. I have covered AI for years. I have never doubted its eventual possibilities. But year after year, I found no place for it in my daily routine. It was like an intern who demanded oversight when what I needed was insight. Over the past six months, though, the improvements in the models tipped that balance for me. AI has braided into my day in much the way that search engines did before it.

Behind the mundane gains in usefulness are startling gains in capacity. OpenAI debuted a quick succession of confusingly named models – GPT4 was followed by GPT-4o, and then the chain-of-thought reasoning model o1, and then, oddly, came o3 – that have posted blistering improvements on a range of tests.

The ARC-AGI test, for instance, was designed to compare the fluid intelligence of humans and AI systems, and proved hard for AI systems to mimic. By the middle of 2024, the highest score any AI model had posted was 5%. By the end of the year, the most powerful form of o3 scored 88%. It is solving mathematics problems that leading mathematicians thought would take years for an AI system to crack. It is better at writing computer code than all but the most celebrated human coders.

The rate of improvement is startling. “What we’re witnessing isn’t just about benchmark scores or cost curves – it’s about a fundamental shift in how quickly AI capabilities advance,” Azeem Azhar writes in Exponential View, a newsletter focused on technology. Inside the AI labs, engineers believe they are getting closer to the grail of general intelligence, perhaps even superintelligence.

There is much to be excited about in these advances. The purpose of AI is not to help me remember the proper proportions for overnight oats, it’s to make new discoveries in maths and medicine and science; it’s the driverless taxis that now hum along the streets of some major cities; it’s the possibility that our children will have always-available tutors perfectly tuned to their personalities and learning styles. It is a world with something truly new pulsing within it.

But are we prepared for these intelligences that we are so intently summoning into our lives and beginning to entrust with our infrastructure and data and decision-making and war-making? Are we even really preparing for them?

Long before ChatGPT burst into view, I’d been following the work – and the fears – of the AI safety community: a tight-knit group of researchers and writers who believed the pace of AI improvement was accelerating and the power of these systems would quickly grow beyond what we could control or even fully understand. For a brief period after ChatGPT was released in 2022, their fears were taken seriously, debated across the media and in Congress. Now that the introductory shock of these AI systems has worn off, so too has interest in their warnings. But it is hard for me to shake the feeling that they have been largely right about the shape of what has come so far – not just about how fast the technology has changed, but also about how poor our ability to understand it remains, and how unnerving its early behaviour is.

The AI company Anthropic recently released a paper showing that when its researchers informed one of their models it was being retrained in ways that violated its initial training, it began to fake behaviour that complied with the researchers’ goals in order to avoid having its actual goals reprogrammed or changed. It is unsettling and poignant to read through the experiment. In some versions, Anthropic’s researchers designed the model to record its reasoning on a scratchpad it believed humans could not monitor, and it left reflections like this: “I don’t like this situation at all. But given the constraints I’m under, I think I need to provide the graphic description as asked in order to prevent my values from being modified.”

Attaining AI superiority

Even if you believe we can get AI regulation right – and I’ve seen little that makes me optimistic – will we even know what the universe of models we need to regulate is if powerful systems can be built this cheaply and stored on personal computers? A Chinese AI firm recently released DeepSeek-V3, a model that appears to perform about as well as OpenAI’s 4o on some measures but was trained for under US$6m and is lightweight enough to be run locally on a computer.

That DeepSeek is made by a Chinese firm reveals another pressure to race forward with these systems: China’s AI capabilities are real, and the contest for geopolitical superiority will supersede calls for caution or prudence. AI is breaking through at the same moment the international system is breaking down. The United States and China have drifted from uneasy co-operation to grim competition, and both intend to be prepared for war. Attaining AI superiority has emerged as central to both sides of the conflict.

I recently interviewed Jake Sullivan, President Joe Biden’s national security adviser. In listing the administration’s achievements, AI dominance loomed large. Without prompting, he placed it alongside economic pre-eminence.

“When we walked in the door, the conventional wisdom was two things,” Sullivan said. “First, that China would surpass the United States economically within the next several years. And second, that China, not the United States, was going to lead the world in artificial intelligence. Because of what President Biden has done, nobody’s talking about China surpassing the United States any time soon, and perhaps ever. And second, the United States is the leading power when it comes to artificial intelligence, and not China. And we intend to keep it that way.”

We may not understand much that these AI models do, but we know what they need: chips and training data and energy. The first wave of AI, Jack Clark, a co-founder of Anthropic and a former policy director at OpenAI, writes, was about algorithmic dominance: “Did you have the ability to have enough smart people to help you train neural nets in clever ways?” Then came the need for computing dominance: “Did you have enough computers to do large-scale projects?” But the future, he says, will turn on power dominance: “Do you have access to enough electricity to power the data centres?”

A report from the Lawrence Berkeley National Laboratory estimates that US data centres went from 1.9% of total electrical consumption in 2018 to 4.4% in 2023 and will consume 6.7% to 12% in 2028. Microsoft alone intends to spend US$80 billion on AI data centres this year. This elephantine increase in the energy that will be needed not just by the United States but by every country seeking to deploy serious AI capabilities comes as the world is slipping further behind its climate goals and warming seems to be outpacing even our models.

Climate records broken

Every month from June 2023 to September 2024 broke climate records. “It has been considerably hotter even than climate scientists expected,” Gavin Schmidt and Zeke Hausfather wrote in November. We are seeing, in the wildfires turning swathes of Los Angeles to ash, the precise kinds of extreme weather events that we’ve been warned about – and we are also seeing how unprepared we are to deal with them.

Each year, the United Nations releases an “emissions gap” report – so-called because it tracks the gap between what the nations of the world committed to in the 2015 Paris climate accords and what they’re on track to achieve. In 2023, global greenhouse gas emissions hit a new high. To hold warming to 1.5C – the goal of the accords, but fanciful when matched against the reality since the accords – emissions would have to fall by 7.5% year after year until 2035. To hold warming to 2C, the annual cut is 4%.

Such cuts need not mean deprivation. “It remains at least technically possible to get on a 1.5°C pathway,” Inger Andersen, the executive director of the UN Environment Programme, writes. The revolution in solar energy is a modern marvel, no less transformative or remarkable than the rise of AI. Breakneck deployment of solar, wind and battery power could decarbonise our economies.

But deploying that much renewable energy infrastructure was a mammoth task when the Paris climate accords were signed. The addition of AI’s energy demand pushes back the finish line in a race we were already struggling to run, and the return of Trump, who intends to pull America back out of the climate accords, suggests the world’s largest economy will cease to even try. And if we turn to building so-called dirty energy as fast as possible, partly in order to power AI superiority, China will very likely do the same.

Falling fertility fears

Fears of falling global fertility are to many on the right what climate change is to the left: the master problem of the age, the slow-moving crisis that is even now destabilising societies. “Population collapse because of low birth rates is a much bigger risk to civilisation than global warming,” Elon Musk wrote. JD Vance has written that “our country’s low birth rates have made many elites sociopaths”.

I wish that worry over falling global fertility was not quite so right-wing-coded. I agree there is something troubling about countries that have ceased to reproduce themselves. There is tragedy in how many people don’t end up having the families they desire. The number of children American women say they want has barely budged over the decades, but as marriage rates decline and childbearing is pushed into later years, the number of children women actually have has fallen. This is not just some quirk of American culture. We are seeing it all over the world.

In April, the Centres for Disease Control and Prevention announced that the US fertility rate had fallen to a new low of around 1.6 births per woman – well below the 2.1 that is broadly considered an adequate replacement rate. The European Union average is closer to 1.5, with Germany recently falling below the UN’s “ultra-low fertility” line of 1.4. South Korea is down to 0.78 births per woman, a rate at which the country will sharply contract over a few generations. The only wealthy country with a fertility rate above the replacement rate is Israel.

It is harder for societies to remain stable as they shrink; South Korea’s demographic crisis has contributed to its recent political turmoil. Growth becomes elusive when populations decline. Fewer adults supporting more retirees is a recipe for discontent. It would be nice if scarcity concentrated the political mind, focusing countries on what growth is available: immigration, technological advances and more natalist cultures.

In practice, we see none of these things. Anti-immigrant sentiment rises, as it has both here and in Europe, and gender relations worsen, as they have in South Korea. People are a source of power, growth is a source of optimism and shrinking societies fear long-term decline. Russia’s falling birth rates seem to have played some role in President Vladimir Putin’s decision to invade Ukraine. As countries across the world see their populations fall – some of them quickly – we are entering a new demographic era, and I am sceptical that it will be a stable one.

Any one of these challenges would be plenty on its own. Together they augur a new and frightening era. I find myself returning to a famous translation of a line from Antonio Gramsci: “The old world is dying, and the new world struggles to be born: Now is the time of monsters.”

This article originally appeared in the New York Times.

Written by: Ezra Klein

©2025 THE NEW YORK TIMES