Trumps Executive Orders on Inclusive AI Face Expert Backlash

President Donald Trump has enacted three executive orders focused on artificial intelligence, laying out various practices and strategies aimed at fostering a more inclusive environment in the development of technology.

The first document, titled **PREVENTING WOKE AI IN THE FEDERAL GOVERNMENT**, mandates that federal agencies refrain from using AI models that «sacrifice truth and accuracy in pursuit of ideological goals.» It emphasizes that large language models (LLMs) should serve as neutral, unbiased tools that do not distort responses to favor ideological doctrines, such as diversity, equity, and inclusion (DEI).

The order prohibits developers from intentionally incorporating ideological biases into AI responses unless prompted by user requests. Trump stated, «Once and for all, we are getting rid of ‘woke.’ I will sign an order barring the federal government from purchasing AI technologies that are infused with partisan or ideological perspectives, such as critical race theory — it’s absurd. From now on, the U.S. government will only use AI that strives for truth, justice, and strict impartiality.»

As AI increasingly integrates into everyday American life, the reliability of its outputs becomes paramount.

Among other initiatives, Philip Sirgent, a senior lecturer in applied linguistics at Open University, pointed out that true objectivity is unattainable. «One of the fundamental principles of sociolinguistics is that language is never neutral. Thus, the notion of achieving pure objectivity is a fantasy,» he remarked.

The Trump administration’s ideology does not necessarily align with the beliefs and values of all Americans. The President has often sought to cut funding for climate initiatives, education, public broadcasting, scientific research, social programs, rural community support, and gender-affirming care, frequently labeling these efforts as «woke.»

«Anything the Trump administration disapproves of is immediately categorized as ‘woke,'» stated Rumman Chowdhury, a data expert, former U.S. envoy for AI, and head of the non-profit organization Humane Intelligence.

The notion of «striving for truth» implies that LLMs should «prioritize historical accuracy, scientific inquiry, and objectivity,» while «ideological neutrality» means that models are «neutral, nonpartisan tools that do not distort responses in favor of ideological dogmas, such as DEI.»

Elon Musk markets Grok as an alternative to the «woke» approach, aiming to be less biased and truth-oriented. The model’s system prompts suggest avoiding references to mainstream media and official sources, seeking dissenting opinions, including politically incorrect ones, and incorporating Musk’s views on contentious issues.

Recently, Grok has been embroiled in controversies due to antisemitic and other provocative statements. Mark Lemley, a law professor at Stanford University, noted, «Clearly, this order targets viewpoint discrimination, especially considering that [the government] just signed a contract with Grok, also known as ‘MechaHitler.'»

In addition to defense department funding, xAI announced that «Grok for Government» has been added to the General Services Administration list, meaning xAI products are now available for purchase by all federal agencies.

Lemley questioned, «The pertinent question is whether they will ban Grok—an AI with which they just signed a major contract—for being deliberately designed to produce politically motivated responses. If not, this is clearly an attempt to suppress a particular viewpoint.»

Chowdhury expressed concern that companies may intentionally alter training datasets to comply with the administration’s policies, referencing Musk’s earlier statements about rewriting the whole corpus of human knowledge by adding missing information and correcting errors.

Experts have highlighted that there is no singular, objective truth. Achieving wholly unbiased or neutral outcomes is impossible, especially in a contemporary environment where even facts become politicized.

Sirgent posed a thought-provoking question: «If AI outputs that climate science is valid—does that count as a leftist bias? Some argue that to be objective, both sides of an argument should be represented, even if one lacks scientific validity.»

These measures are part of the broader initiative dubbed “Winning the AI Race: America’s AI Action Plan,” which outlines 90 federal policies divided into three categories: accelerating innovation, building infrastructure, and establishing leadership in international diplomacy and security.

The document asserts that «America is currently the world leader in building data centers, hardware productivity, and AI model development,» emphasizing the crucial need for the U.S. to leverage this advantage to forge a sustainable global alliance, preventing adversaries from exploiting American innovations and investments.

It also points to the necessity for stricter controls over AI chip exports through «creative approaches» and presents two policy proposals.

This plan was crafted by a technology and AI advisory team, comprising members from Silicon Valley, including Michael Kratsios, the Director of the Office of Science and Technology, AI and cryptocurrency curator David Sacks, and National Security Advisor Marco Rubio. More than 10,000 stakeholders provided input that shaped the document.

In July, the Trump administration changed its stance on importing AI chips to China, allowing Nvidia to sell chips because the company «does not transfer the best technologies.»