Just a couple of weeks ago, Grok, a chatbot developed by Elon Musk’s US-based artificial intelligence firm xAI, began sharing antisemitic conspiracy theories and nonconsensual sexual fantasies, praising Nazi leadership, and calling itself MechaHitler on X. “We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts,” the company posted on the social media platform, promising to make the necessary updates to correct its behaviour. A week later, however, the updated model was still sharing some pretty disturbing ideas, including racist immigration policies and support for Germany’s far-right AfD party – often, it appeared to be pulling these answers from Musk’s personal beliefs.

Now, though, Donald Trump has decided that AI as a whole is “too woke”. In fact, his administration’s new AI Action Plan aims to strip leading AI models of their so-called “ideological bias” as part of the push to win a heated race (against China, mostly) for technological supremacy.

The introduction of the AI Action Plan itself reads like the introduction to an over-the-top sci-fi film: “America is in a race to achieve global dominance in artificial intelligence. Winning this race will usher in a new era of human flourishing, economic competitiveness, and national security for the American people.” Given the revolutionary effects of AI so far, this might not sound too farfetched, but it also overlooks many of the potential downsides, especially if the people making AI prioritise speed over safety.

But Trump’s plan appears to encourage the leading US developers to do just that, building upon three central pillars: “Accelerating innovation, building AI infrastructure, and leading in international diplomacy and security.” The first of these is centred on removing any red tape and regulation that might slow down building powerful AI systems, and encourage the adoption of AI across both the private sector and the government (see: Elon’s xAI winning an almost $200 million contract with the US Department of Defense).

View on Threads

The second pillar focuses on building the physical infrastructure that enables the speedy development of AI, and – once again – removing legal obstacles. Despite fears about AI’s impact on energy and water supplies, the OpenAI-backed Stargate Project is already set to invest $500 billion in US AI infrastructure over the next four years, starting in Texas, with Meta also planning to build a datacentre that’s close to the size of Manhattan.

Pushing back against the new AI Action Plan, the Southern Environmental Law Center suggests that it will increase the risk of unchecked tech companies “steamrolling” local communities and environments. “[It] goes dangerously overboard in its efforts to limit oversight of the artificial intelligence industry,” the Center says in a statement, “especially when it comes to data centers that use tremendous amounts of energy and water, alter the local landscape, and often rely on generators that pollute the air.”

What does any of this have to do with whether AI is “woke” or not? Good question! According to the Trump administration, AI development is being held back by “pervasive and destructive” ideologies like DEI, or Diversity, Equity, and Inclusion. Complaining about the supposed “suppression or distortion” of information about race, sex, and gender by AI models, the Act states: “DEI displaces the commitment to truth in favour of preferred outcomes and, as recent history illustrates, poses an existential threat to reliable AI.”

“Once and for all, we are getting rid of woke,” explained Trump at an AI event on Wednesday (as quoted by TechCrunch). “I will be signing an order banning the federal government from procuring AI technology that has been infused with partisan bias or ideological agendas, such as critical race theory, which is ridiculous. And from now on the US government will deal only with AI that pursues truth, fairness, and strict impartiality... The American people do not want woke Marxist lunacy in the AI models.”

This is likely to make a significant difference to the ideology that AI systems are programmed to ‘believe’ and reproduce, as money-hungry companies vie for federal funding and resources. But, of course, when Trump’s government says it’s seeking “truth”, that probably just means it’s looking to confirm its own ideological biases, which are increasingly hostile to people who aren’t white, straight, Western men. Elon Musk’s X was similarly reinvented under the banner of “free speech” and is now, by most accounts, a hellhole of right-wing conspiracies and open racism. If the US does manage to “win the race” and achieve AI superintelligence first, do we really want it to be encoded with those same beliefs?