United States, explained.

Make America Great Again caps in the shape of a computer (Imagen/Canva)

Behaviour-predicting AI has advanced rapidly since Trump was president (Imagen/Canva)

AI might better predict Trump’s foreign policy

A method marked the madness and can be decoded.

Ben Scott

10 September 20244 min read

If Trump’s first-term foreign policy was unpredictable, it’s not because his worldview was unknown. He’d been telling us about it since the 1980s and his views haven’t changed much since in that time.

Trump understands America’s relationships with the rest of the world only in bilateral, transactional and, usually, personal terms. He reduces America’s economic relationships to balance sheets of bilateral trade surpluses (good) and deficits (bad). Trump loves tariffs because they both impede exports and raise revenue. Denying the consensus, he claims that they’re “a tax that doesn’t affect our country”.

Trump sees America’s security relationships only as a service for customers. He hates “freeloaders”, which is how he views many of America’s allies and partners. In February, Trump told Russia to do “whatever the hell they want” to any NATO member country that wasn’t contributing enough.

But even fully paid-up customers have reason to doubt that Trump would defend them. Riyadh invested heavily in its relationship with the Trump administration but was disappointed that the United States did not respond to a September 2019 Iranian attack on Saudi oil facilities that temporarily cut global oil production by five per cent. More recently, Trump has expressed reluctance to defend Taiwan, which has on order more than $14 billion worth of US military equipment, because it “doesn’t give us anything”.

Although Trump liked having a powerful military, he was leery of using it. He prefers to make deals with America’s adversaries, even though he’s very bad at it.

Although we know Trump’s world view, it has been harder to anticipate how he will apply it.

Trump’s pursuit of personal relationships with “strongmen” such as Russian President Vladimir Putin perplexed his national security advisors HR McMaster and John Bolton. It shouldn’t have. His behaviour has been consistent and accords closely with his reductive worldview.

It’s true that Trump’s praise for President Xi Jinping faded as he came to blame China for his political woes, because of Covid-19. But he’s since resumed lauding Xi as “smart, brilliant, everything perfect”. His enthusiasm for a deal with Xi caught Trump’s national security team off guard. They had convinced Trump that Beijing interfered against Republican candidates in the November 2018 mid-terms. But Trump did not reprimand Xi when they dined in Buenos Aires the following month. Instead, he begged Xi for help with his own re-election.

Authoritarian regimes have correctly predicted Trump’s behaviour. North Korea’s success in securing Trump’s agreement to unconditional meetings with Kim Jong-un is a striking example. In May 2018, Pyongyang’s envoy – the highly-sanctioned general Kim Yong-chol – talked his way into the oval office, against the objections of Trump’s staff. He delivered the very large letter that started the Trump-Kim bromance.

Although we know Trump’s worldview, it has been harder to anticipate how he will apply it. That’s because Trump is typically guided by what he thinks will make him look best in the moment, and often influenced by the last person he spoke to. He cancelled 2019 airstrikes on Iran in retaliation for shooting down a US Global Hawk drone after the White House lawyer told him they could cause 150 casualties. Accounts of his far more risk-accepting 2020 decision to kill Qassem Soleimani, head of Iran’s Islamic Revolutionary Guards’ Quds Force, are “differing and incomplete” but some insiders said Trump was motivated by “negative coverage after his 2019 decision to call off the airstrikes …[and]… felt he looked weak.”

Some observers even see method in madness, arguing that Trump had adopted the “madman theory” of foreign policy, popularised by Richard Nixon, which aims to keep America’s opponents off balance. But it’s more likely that America’s adversaries have been learning from Trump’s first term how he might behave if he returns for a second.

Trump back in office may be less constrained than the first time. But he should be more predictable. There is an extraordinary amount of information about Trump’s presidential decision-making. Information security was not a high priority in his White House. The available data ranges from his impulsive tweet storms through to a series of books written by former staff. Access to Trump’s phone data – or his television remote – would provide even more insight.

Artificial intelligence might make the most sense of that data. Behaviour-predicting AI has advanced rapidly since Trump was president. For example, researchers at MIT and the University of Washington have recently developed a new way to account for “suboptimal” human decision-making in their behaviour-predicting algorithm.

With Trump’s return to the White House a realistic possibility, the incentives for governments – and corporations – to invest in such a capability are high.

About the author

Ben Scott

Ben Scott is a Senior Advisor at the Australian National University’s National Security College.

Topics

United States Elections