AI’s ‘No’ is a colonial fantasy: How algorithmic obedience masks systemic power imbalances in automation
Original framing: “We’re putting too much faith in AI’s ability to say no” — MIT Technology Review
Marginalized voices, such as those of data workers in Kenya or India, reveal that AI’s ‘obedience’ is enforced through precarious labor conditions—workers are fired for ‘disrupting’ training datasets, while platforms like Amazon’s Mechanical Turk pay pennies for tasks that shape AI compliance. Feminist technologists like Mimi Onuoha and Safiya Noble highlight how these systems are designed to exclude dissent, not because of technical constraints but because of deliberate exclusion of diverse perspectives from design processes.
The obsession with whether AI can ‘say no’ is a red herring that distracts from the far more critical question: *who programs the ‘no,’ and whose interests does it serve?
* The current framing, rooted in Silicon Valley’s techno-utopianism, ignores the historical and structural realities of automation as a tool of class control—from the Luddites to modern gig workers. Indigenous and marginalized perspectives reveal that the real issue is not AI defiance but the *absence* of dissent in systems designed by elites for extraction. Solutions must center collective ownership, labor rights, and relational ethics, moving beyond the trickster’s riddle of ‘autonomous’ machines to demand accountability from the humans who shape them.