Not Neutral: A Critical Guide to AI

Artificial intelligence (AI) isn't new: the term was coined in 1956 and the ideas behind it stretch back even earlier, to mid-20th century cybernetics. Over the past decade, however, AI has taken over public discourse and seeped into almost every part of our lives, from education and health care to immigration control and social media. And yet, rather than the messianic salvation Big Tech promises, the world around us is hurtling toward collapse: the environment is on the brink of destruction, social programs and safety nets are being dismantled, legal protections for workers, queer, nonbinary and trans people, Indigenous peoples, communities of color, women, immigrants and other minoritized groups are under attack, and fascist ideologies are gaining ground. In this reality, the hype around AI: all the promises of innovation, efficiency, and convenience, rings hollow. It's clear that AI isn’t here to liberate us; it’s here to cement and amplify the systems that harm us.

AI is not neutral:

• AI is not identity-neutral, despite Big Tech’s claims that it is "color-blind." By pretending that context, including bodies, histories, and relations of power, doesn’t matter, AI systems reify difference and reinforce structural violence. Biometric profiling, predictive policing, and algorithmic injustice are not "glitches": they are built into the AI systems themselves.
• AI is not spatially neutral either. It does not impact everyone in the same way, everywhere: most AI systems are trained predominantly on English-language data and reflect the norms, biases, and assumptions of the Global North. They perform far less accurately, and often far more harmfully, across different geographic, linguistic, ethnic, and cultural contexts. AI harms disproportionately target the Global South, Indigenous communities, and marginalized populations, and therefore its effects must be studied in situated ways.
• AI is most definitely not carbon-neutral. The fantasy of AI as a disembodied “cloud” service Big Tech is selling us deliberately erases the transnational labor, material extraction, and political infrastructures that make AI possible. Training and deploying models like ChatGPT requires enormous energy, massive amounts of resources, and sprawling networks of data centers, often marketed as economic revitalization strategies. In reality, these infrastructures devastate ecosystems, displace frontline communities, and further entrench environmental racism.

This guide doesn’t pretend to offer a "solution" to the multiple crises intensified by AI - this techno-solutionist mindset is part of the very logic that got us here in the first place. Instead, it’s a resource for collective study and action, rooted in decolonial, abolitionist, and queerfeminist commitments mobilized against what we call the expanding global order of AI Empire: a network of technologies, infrastructures, ideologies, and power relations that depend on ongoing extraction, exploitation, and exclusion. Mainstream computer science and tech industries have every interest in keeping these systems running and expanding. That’s why we insist on a different lens which not only asks how AI might be “reformed,” but questions whether it should exist in its current form at all. To fully understand the impact of AI, we have to go deeper than surface-level conversations about “bias.” "Bias," "accuracy," and "privacy" are just the tip of the AI Empire iceberg. Underneath are the systems that AI helps uphold:

• Labor exploitation: AI depends on ghost work - the hidden, underpaid, and dehumanizing labor that keeps the illusion of "automation" alive, especially in the Global South.
• Environmental destruction: Training and deploying AI models requires enormous amounts of energy and relies on violent extraction of rare earth minerals, which destroys local ecosystems, displaces and dispossesses communities, and worsens their quality of life.
• Distributive, epistemic, and representational violence: AI technologies redistribute power and resources upward, while locking in racialized, gendered, and classed inequalities, and manufacturing new oppressive narratives about who matters and who is disposable.

The goal of this guide is not just to critique, but to spark imagination, action, and transformation. We believe technology should be in service of collective liberation, not domination.
We hope this guide helps you study, organize, and work toward a different world, rooted in justice, care, reciprocity, and solidarity.
 

This guide was collaboratively developed by Sarah Appedu, a PhD student at the iSchool; Brenna Helmstutler, iSchool and Sport Industry Librarian; and Jasmina Tacheva, an Assistant Professor at the iSchool.

You can travel through this guide by theme, depending on what calls you:

• AI Hype vs. Reality: Understand the myths the AI industry sells us (progress, objectivity, inevitability) and why they're harmful
• Intermeshing Systems of Harm: Explore the colonial profit-driven infrastructures at the heart of the complex ecosystems of AI, including the politics of classification, calculation, datafication, labor extraction, environmental destruction, racial and gender violence
• Abolitionist Resistance and Refusal: Learn about the people and coalitions seeking to abolish the logics of AI Empire 
• Tools for Action: Find tangible steps each of us can take to disrupt and mitigate the harms of AI
• Critical AI Pedagogies: Tips and resources for teaching, organizing, and building abolitionist consciousness