The Problem with Intelligence: Deconstructing the Ideological Ghost in the AI Machine

TL;DR

"Intelligence" is not a neutral metric. Stephen Cave’s seminal work argues that the concept is a "thick" value-laden term born from a history of patriarchy, colonialism, and scientific racism. By fetishizing intelligence as the ultimate measure of worth, the AI industry risks perpetuating ancient dominance hierarchies under the banner of high-tech innovation.

Contextual Positioning

While most AI ethics papers focus on "de-biasing" datasets, Stephen Cave’s work—presented at the 2020 AAAI/ACM Conference on AI, Ethics, and Society—operates at a more fundamental level. It is a philosophical deconstruction of the very foundation of Artificial Intelligence. Cave suggests that the field's obsession with "intelligence" is not a pursuit of pure science, but a continuation of an ideological project that has historically used "mental ability" to justify the power of one group over another.

The "Matrix of Domination": A Value-Laden History

The paper argues that intelligence has consistently served as a tool for establishing social and political hierarchies.

• The Aristotelian Root: Aristotle argued that those with "reason" (men, the elite) were born to rule, while those lacking it were "slaves by nature."
• The Colonial Logic: In the 18th and 19th centuries, European powers used "superior intellect" and technology to justify the conquest of "savage" nature and peoples—the "White Man's Burden."
• The Eugenicist Turn: The term "intelligence" rose to prominence alongside the eugenics movement. Figures like Francis Galton and Lewis Terman (creator of the Stanford-Binet IQ test) sought to quantify mental ability to reinforce racial and class segregation.

Figure 1: Title and institutional context of the AI ethics inquiry.

Methodology: How Ideology Shapes the Future of AI

Cave identifies five critical ways this "value-laden" history impacts we perceive AI today:

1. The Fetishization of Intelligence

Prominent figures like Stephen Hawking and Elon Musk often claim that "everything civilization offers is a product of intelligence." Cave argues this is empirically false. It ignores virtues like altruism, cool-headedness, and social adroitness, choosing instead to elevate a trait historically associated with the white male elite.

2. The Lack of Diversity (The "Brilliance" Myth)

Industries that prize "raw brilliance" over effort tend to have fewer women and people of color. The AI sector’s "meritocracy" myth mask a bias: when a culture values "pure genius," it implicitly defaults to the historical image of the genius—white and male.

3. Mastery of Nature vs. Climate Reality

We pitch AI as the "solution" to climate change, a narrative that continues the Enlightenment goal of total technological mastery over nature. Cave warns this might be a "moral hazard," masking the need for behavioral change with a palliative belief in a "smarter" tool.

4. Fear of the Superintelligent Master

Why does the West fear the "Terminator" or a superintelligent enslaver? Cave’s insight is profound: because for centuries, we have used the "intellectual superiority = right to rule" logic to enslave others. We fear that a more intelligent machine will simply treat us the way we have treated those we deemed "inferior."

Figure 2: The paper serves as a "consciousness-raising" act for the AI community.

Critical Analysis & Takeaways

The paper provides a necessary "consciousness-raising" for the AI community. Its primary contribution is challenging the Inductive Bias of the entire field: the assumption that building "more intelligence" is a self-evident good.

Key Limitations: The paper does not offer an alternative metric for AI success, nor does it dive into the technical feasibility of "decolonizing" an algorithm. It functions as a diagnostic tool rather than a technical manual.

Looking Forward: As we move toward AGI, Cave’s work suggests that our greatest challenge isn't just an "alignment problem"—the risk that AI won't do what we want—but an identity problem. If our definition of intelligence is inextricably linked to dominance, then any "superintelligence" we build will inevitably reflect the darker chapters of our own history.

Conclusion

We must move beyond the "thick" concept of intelligence to a broader understanding of human flourishing. Only by recognizing the ideological baggage of our terms can we build an AI that serves humanity rather than just the latest iteration of a dominance hierarchy.