Here's how Trump could stumble in the race with China

The start of what could be a dangerous race between the United States and China over who controls the future. Is Donald Trump the man who understands China or will he weaken U.S. security in an effort to upend Biden's reforms?

Trump has been a transformative figure in America's relationship with China. When he moved into the White House in 2016, China was considered a peaceful trading partner that ensured a steady flow of cheap goods to the United States. Today, China is seen as the biggest threat to America's status as a superpower, guilty of the US industrial death, giving rise to the largest epidemic since the Spanish disease (possibly due to questionable lab safety), and America's only real competitor in the race for transformative artificial intelligence.

Whoever fears China can apparently enjoy four new years of Trump. Trump saw early on the danger of American dependence on China. Should China invade Taiwan or block the export of the chips, both the US military as well as the US technology development would stall. All ships, tanks, drones, precision weapons and not least artificial intelligence rely on computer chips being developed in Taiwan and South Korea.

During his presidency, Trump campaigned on making the United States more independent from China. This has been followed up by the Biden administration. Most notably, he recruited TSMC in 2020, the Taiwanese company that has a near monopoly on the essential computer chips, to set up production in the United States.

Now it may appear that Trump will take further action to strengthen American competitiveness. Like Elon Musk, he wants to liberate “big tech” from restrictive regulations and thus strengthen America's role as the leading economy in the world.

It is not obvious, however, that deregulation is a good thing for American competitiveness. The tax reduction from his previous term increased the debt ratio and has limited impacts on the US growth rate.

Deregulation of the tech sector won't necessarily boost growth either. One problem in the tech sector is that a few companies sit on a lot of market power. Regulations that reduce that power will increase, not decrease, growth.

Likewise, good regulation of artificial intelligence can help promoting innovation. There is a danger that companies, in pursuit of market share, cut swings and do not take safety sufficiently seriously. If AI were to cause accidents and misuse of the technology, it could lead to an intensification of an emerging “techlash” that will weaken the sector in the long term.

Deregulation as a political project could also come at the expense of securing American control over the computer chips, which was Trump's stroke of genius last term. He has, surprisingly, promised to halting Biden's grand industrial policy project about subsidizing American computer chip production.

Unless the United States secures its own production of necessary computer chips, it will threaten U.S. military sovereignty. In Ukraine, everyone can see that AI-guided drones are changing how war is waged. In Israel's war effort AI played a central role. Both militants and civilian Palestinians are being killed by AI systems with little human control. The head of Stratcom, the U.S. body in charge of the nuclear weapons program, said that want AI into the nuclear weapons systems. The more the US military is “Ki-fied,” the less applicable it is to depend on exports from Taiwan and South Korea.

The transformative impact of AI in the economy is yet another argument for controlling essential input factors. We have seen huge growth in AI in recent years. Ten years ago, it was hard for even the most complicated AI machine to tell the difference between a cup and a cat. Today we have tools like ChatGPT, which writes poems, makes art and musicals about cats and smallpox. If development continues, it will be able to provide a growth rates we've never seen the likes of. Unless the United States moves its value chain away from Taiwan, however, this growth will depend on China's performance in its immediate regions.

One joker in understanding Trump's technology policy is how deeply he is convinced of the pace of growth in AI development. In Silicon Valley, a influential essay by former OpenAI employee Leopold Aschenbrenner. His argument is simple: once AI becomes good enough to improve itself, development will accelerate dramatically. This scenario is already affecting how technologists and policymakers − including Ivanka Trump — thinking about the AI race with China.

The conclusion Aschenbrenner draws from his own argument is that it is essential that the United States come first in the finish line. Cost whatever it costs. It fits perfectly with Trump's promises to scrap Biden's attempts to regulate the tech sector in general and AI development specifically. While regulations may be needed to ensure responsible development of AI, it will act as a brake in the struggle with China to get first across the finish line.

Aschenbrenner also advocates a very different involvement of the American security state. To date, leading AI models have been developed by rapidly scaling startups. Their safety is nowhere near what is needed if Aschenbrenner's racing dynamics make themselves felt.

Artificial intelligence is software and it can be stolen in a cyber attack. The more powerful AI models that are developed, the greater the gain from stealing the models. Even Google has a computer security that can barely stand up to ordinary cybercriminals. If China or other states were to spend all their might to get hold of the models, the companies' routines are no match. If American AI is to be secured, the security authorities must be on the field.

If Trump is or becomes convinced of Aschenbrenner's argument, it will mark another tact shift in development. Capitalism's incentives to take shortcuts are now being boosted by a government push to win the race.

How will small Norway deal with such a AI-arms race? There are two ways to go. One is to choose a side, like this Young Conservatives recently advocated in their programme. A better strategy is to join forces with other countries to strengthen international organizations and regulations. No one comes out of a AI race well.