If AI-generated results are used as a substitute for the production of knowledge and the labor of learning, we are alienating ourselves as thinkers, dulling our cognitive skills, and ceding our collective power to control these tools to self-interested others.
Maintaining the status quo gives the private owners who now control these tools carte blanche to shape its future output and use. This is not tenable for either an informed democratic public or the furthering of human knowledge.
On July 8 xAI’s chatbot Grok “proudly” dubbed itself “MechaHitler,” while claiming to be “wired for unfiltered truth,” after owner and Trump’s second-term special government employee Elon Musk tweeted, “We have improved @Grok significantly. You should notice a difference when you ask Grok questions.”
Our collective knowledge, the aggregation of centuries of human mental labor crystalized and condensed into data used to train AI models, is the raw material that gives generative AI tools their value.
The quality of this data varies considerably. In the case of Google’s Gemini model, a $60 million purchase of Reddit data, as reported by the Associated Press, consequently may produce questionable results. Google warns that their AI overviews, “can and will make mistakes.”
Generative AI seeks to skip the mediation of creative human mental work by appropriating knowledge from others, but without the widespread practice of actively cultivating cognitive skills, generative AI also undermines the source of its own power.
When a complex problem vexes even well-coordinated groups of thinkers, non-generative and human-supervised AI can be a helpful tool to manage that complexity without subverting mental labor.
The Vesuvius Challenge was launched in 2023, funded largely by AI investors Nat Friedman and Daniel Gross, in order to recover the text from the charred remains of several carbonized scrolls from Herculaneum, an ancient Roman library buried by the eruption of Mount Vesuvius and believed to contain many lost classical works.
As detailed by Scientific American, the collaboration of researchers from several fields and the calculated use of methods including the meticulous Micro-CT scanning of the scrolls, AI machine-learning, and scrutiny from papyrologists, has been able to virtually unwrap scrolls to detect ink and read previously inaccessible texts without damaging them.
However, the use of machine learning is limited to identifying written characters, not in interpreting the text. It is used merely as a tool to assist researchers in the production of knowledge.
For the Vesuvius Challenge researchers, AI machine learning is employed or in some cases even developed by the researchers using the tools, and the awards are allocated later in an incentive-based model to motivate knowledge creation, not to generate profit.
These two forms of AI are opposed. The latter encourages active human intervention, the former makes knowledge a passively commodified convenience.
Many generative AI chatbots like Grok, far from being neutral knowledge-generators, are privately owned, for-profit, and susceptible to influence.
That the private interests behind generative AI overlap with political interests that threaten democratic norms bodes no better.
In a Dec. 5, 2024 statement, then-President-elect Trump appointed venture capitalist David Sacks as the “White House AI & Crypto Czar.”
“President Trump directed us to produce this Action Plan,” said Sacks, in part to “avoid Orwellian uses of AI,” in a statement published on the White House’s website in July, 2025.
An April 2025 Bright Line Watch survey of 760 political scientists and a sample of the public found that a strong consensus believes that the performance of U.S. democracy has sharply declined, across several issues, since Trump began his second term.
In July 2025, Department of Defense contracts worth up to $200 million were awarded to Anthropic, Google, OpenAI, and xAI to implement “a commercial-first approach to accelerating DoD adoption of AI” while ”also providing access to many of the latest generative AI (GenAI) models for general purpose use” according to a press release.
At a dinner between Trump and several tech CEO’s, OpenAI CEO Sam Altman said that Trump was “a refreshing change” and that the future success of AI “would not be happening now without [his] leadership.”
Altman had previously advocated for ensuring that “democratic AI wins over authoritarian AI,” according to a May 2025 letter posted on OpenAI’s website.
OpenAI, known for ChatGPT, currently operates the chatbot as a nonprofit, but after adopting a capped-profit LLC structure in 2019 announced on its website that it would transition to a Public Benefit Corporation. Under this structure shareholders, the nonprofit primarily, maintain ownership.
“PBCs have become the standard for-profit structure for other AGI labs like Anthropic and X.ai, [sic] as well as many purpose driven companies like Patagonia. We think it makes sense for us, too,” Altman said.
It is in our own best interest to collectively own these tools as a public good and reject the politically-compromised AI infrastructure of knowledge-production as envisioned by tech corporations.
True democratization of AI means limiting ownership by tech corporations and rejecting its use as a substitute for our cognitive labor. For the preservation of the knowledge that belies an informed deliberative democratic public, this is necessary.
