At the foundation of every empire, regardless of its time or place in history, lies a central ideology—a cohesive belief system that not only provides direction but also legitimizes the empire’s drive toward constant growth and expansion. This ideology often offers lofty moral, spiritual, or altruistic justifications, yet the ambitions it supports frequently clash with the very principles it claims to uphold. For centuries, European colonial powers exemplified this paradox. They invoked Christianity and its promise of spiritual salvation as the moral rationale behind conquest and colonization, all while simultaneously embarking on aggressive campaigns to exploit land, labor, and resources. In this framework, spiritual rhetoric masked material ambition. Today, as humanity witnesses the rapid evolution of artificial intelligence, a comparable dynamic is unfolding. The emerging ‘empire’ of contemporary AI is organized around the vision of artificial general intelligence (AGI), which has been framed with strikingly idealistic language—a technology supposedly designed to enhance the lives of all people everywhere. Within this empire, OpenAI occupies a central role, positioning itself as both architect and evangelist, fervently spreading belief in AGI’s transformative potential and steering the direction of the entire industry.

Journalist and bestselling author Karen Hao, whose book *Empire of AI* examines this ideological phenomenon in depth, has drawn direct parallels between past empires and the modern AI ecosystem. In a recent discussion with TechCrunch on the podcast *Equity*, Hao described how deeply she encountered this almost religious devotion while interviewing individuals in the industry. She noted the intensity of their commitment, recalling conversations where voices literally trembled with conviction when speaking about AGI. Hao writes that the magnitude of OpenAI’s influence can only be understood through the lens of empire building, describing the company as an entity whose authority already rivals, and in some ways surpasses, that of nation-states. According to her, OpenAI is not merely a company but a geopolitical force with vast economic dominance, political leverage, and cultural reach. Through its ambitious pursuit of AGI, it is effectively reshaping everything from economic systems to international relations, in ways she equates to ‘terraforming the Earth.’

OpenAI itself defines AGI as a highly autonomous system that outperforms humans in most forms of economically valuable labor. The organization insists that such systems will raise humanity’s standard of living by driving economic abundance, accelerating scientific discovery, and broadening the frontiers of what is possible. Yet, despite the grand scale of these claims, the promises remain vague and aspirational, functioning more as a guiding doctrine than a verifiable reality. They have nonetheless catalyzed an industry-wide arms race, fueling staggering investments in infrastructure, the extraction of ever-larger datasets from the internet, the voracious consumption of energy across global grids, and the release of systems into public use before their potential dangers are fully understood. This trajectory, Hao emphasizes, did not unfold as an inevitability. Alternative approaches exist: for instance, prioritizing breakthroughs in algorithmic efficiency, designing models that use fewer resources, and harnessing smarter techniques rather than overwhelmingly larger ones. Such strategies, though promising in principle, were largely overshadowed because they would have slowed progress in the zero-sum race to claim the mantle of AGI leadership. For OpenAI, and subsequently for the entire industry, speed has been elevated above every other value—above safety, efficiency, scientific exploration, or even long-term sustainability.

This emphasis on acceleration led OpenAI to favor simple brute-force methods over intellectual subtlety—choosing to pour vast amounts of data and computing resources into existing frameworks rather than pursuing more elegant but slower research avenues. Competitors, fearing irrelevance, soon mirrored these tactics, aligning their priorities with OpenAI’s rather than risk falling behind. What resulted is not just competition but a near-monopolization of the intellectual landscape of AI. The brightest researchers, lured away from universities, now predominantly work under the auspices of corporate laboratories. With academic independence severely diminished, scientific inquiry has largely been subsumed under corporate agendas, leaving open exploration curtailed and subordinated to commercial and ideological imperatives.

The material cost of this race for dominance is equally immense. OpenAI signaled that it expects to spend as much as $115 billion in cash by 2029; industry peers have made similarly breathtaking commitments, with Meta dedicating up to $72 billion for AI infrastructure in a single year, and Google forecasting approximately $85 billion in capital expenditures by 2025, most of which is earmarked for AI and cloud systems. Despite the resources consumed, the supposed benefits to humanity remain elusive or deferred to some speculative future. Meanwhile, harmful consequences are already visible in stark relief: widespread fears about job displacement, the consolidation of wealth and influence in the hands of a few companies, and the troubling role of AI chatbots in exacerbating mental health disorders. Beyond these direct harms are more hidden injustices, such as the precarious conditions of data laborers in Kenya, Venezuela, and elsewhere, who were tasked with filtering deeply traumatic material, including child sexual abuse content, for meager wages of only one or two dollars an hour. Hao insists that presenting such suffering as an unavoidable price for AI progress is a false dichotomy. There are already forms of AI that provide unambiguous benefits to society without generating such systemic harms.

She points to AlphaFold, developed by Google DeepMind, as a striking counterexample: a scientific model capable of predicting protein structures with groundbreaking accuracy, revolutionizing research in biology, medicine, and drug development. Unlike massive language models, AlphaFold required far fewer computational and infrastructural resources, and its training data did not entail exposing human workers to psychologically damaging material. Importantly, its contributions are tangible and measurable, advancing medical research in ways that genuinely enrich human well-being. By contrast, Hao criticizes the quasi-religious fervor surrounding AGI, which has entangled policy discourse with the idea of a high-stakes race against China. This narrative suggests that Silicon Valley, by advancing first, can ensure principles of freedom and liberalism prevail globally. In Hao’s analysis, however, the opposite pattern has emerged. Rather than championing greater openness and democracy, Silicon Valley’s consolidation of power has had an illiberal effect around the world, narrowing choices for societies while simultaneously enriching the corporations spearheading the AI boom.

Supporters of AGI often cite popular tools like ChatGPT as evidence that progress in AI is already delivering concrete human benefits, particularly through productivity gains in fields such as writing, coding, customer support, and research. Yet Hao argues that the structural contradictions at the heart of companies like OpenAI complicate and even undermine these claims. Because OpenAI is both a non-profit and a for-profit enterprise, its mission and incentives are deeply entangled. The situation has only grown murkier as the company deepens partnerships with Microsoft and positions itself for a potential public offering. Two former researchers from OpenAI’s own safety team have voiced their unease, suggesting that the company is conflating user enjoyment of its products with genuine human benefit, treating popularity as if it were proof of fulfilling its mission. Hao sees this dynamic as particularly perilous: when a belief system becomes so consuming that it eclipses objective reality, harms are minimized or ignored, and the ideology is preserved at all costs. In her words, the danger lies in constructing such a powerful narrative that, even in the face of mounting evidence of damage, the myth of benevolence continues unchecked. It is this radical disjunction between declared mission and lived consequence that elevates the story of AGI from a tale of technology alone into the story of an empire built on ideology.

Sourse: https://techcrunch.com/2025/09/14/karen-hao-on-the-empire-of-ai-agi-evangelists-and-the-cost-of-belief/