Trump Targets ‘Woke’ Ideals in New AI Executive Order
President Donald Trump has signed an executive order aimed at ensuring that artificial intelligence (AI) technologies acquired by the federal government remain free from what he terms “woke” ideologies. Announced on October 25, 2023, this directive prohibits the government from procuring AI technologies that contain “partisan bias or ideological agendas,” including concepts such as critical race theory.
The executive order is part of the White House’s broader AI action plan, which outlines various initiatives and policy recommendations intended to position the United States as a leader in AI development. This latest move signifies a notable expansion of Trump’s campaign against diversity, equity, and inclusion (DEI) principles within the tech sector.
The order mandates that large language models (LLMs), which power advanced chatbots like ChatGPT, must adhere to two main principles: “truth-seeking” and “ideological neutrality.” According to Trump, “From now on, the US government will deal only with AI that pursues truth, fairness, and strict impartiality.” This directive raises important questions about the nature of bias in AI systems.
Understanding AI Bias and Ideology
Experts contend that whether AI can be categorized as ideologically biased is not a straightforward matter. AI models primarily reflect the data on which they are trained, the feedback they receive, and the specific instructions provided during their development. This complexity means that responses from AI chatbots can sometimes appear “woke,” a term that lacks a precise definition and can vary in interpretation.
Oren Etzioni, former CEO of the Allen Institute for Artificial Intelligence, explained, “AI models don’t have beliefs or biases the way that people do, but they can exhibit biases or systematic leanings, particularly in response to certain queries.” This perspective underscores the ongoing debate surrounding the presence of bias in AI technologies.
The executive order specifically delineates that LLMs utilized in government applications should prioritize historical accuracy and scientific inquiry. Furthermore, it calls for developers to avoid embedding partisan or ideological judgments into AI responses unless explicitly prompted by users.
The order emphasizes a focus on government procurement of AI technologies, stating that the federal government should approach regulating the private AI marketplace with caution. Notably, major tech firms such as Google, OpenAI, Anthropic, and xAI have recently secured substantial contracts aimed at enhancing the Department of Defense’s adoption of advanced AI capabilities.
Political Backlash and Implications
Trump’s actions reflect a longstanding narrative of perceived bias within the tech industry. In 2019, his administration encouraged users to report instances of perceived censorship on platforms like Twitter and Facebook. Notably, a 2020 Facebook study revealed that conservative news content often garnered significantly more engagement than neutral content on the platform.
In response to the executive order, Senator Edward Markey (D-Massachusetts) expressed concerns about the implications of Trump’s “anti-woke AI actions.” He sent letters to multiple tech CEOs, arguing that the Republican party’s effort to leverage political influence to reshape platforms’ speech is both dangerous and unconstitutional.
Experts also point out that while bias interpretations may differ among users, there is evidence suggesting that certain AI models lean towards leftist viewpoints. Research from the Stanford Graduate School of Business found that many Americans perceive popular AI models as exhibiting a leftward slant.
Andrew Hall, a Stanford professor involved in the research, noted, “There’s definitely evidence that, by default, when they’re not personalized to you, the models on average take left-wing positions.” This phenomenon could stem from the methods companies employ to ensure their chatbots avoid producing content deemed offensive, which may inadvertently lead to skewed outputs.
Challenges arise in defining and enforcing new policies regarding ideological bias in AI. The executive order suggests that vendors should disclose their model’s system prompts—guidelines that dictate how AI responds to queries—as part of compliance with the new requirements. However, questions persist regarding how adherence to these principles will be evaluated.
Trump’s directive may introduce additional obstacles for AI companies seeking government contracts, potentially hindering innovation in a sector that thrives on rapid development. Etzioni expressed concern that such regulations could create complexities for developers, stating, “This type of thing creates all kinds of concerns and liability for the people developing these models.”
As the conversation about AI bias continues, the implications of Trump’s executive order will likely reverberate through the technology landscape, influencing not only government procurement but also the broader discourse surrounding AI ethics and governance.
