Beyond Digital Literacy: Toward Structural Cognitive Sovereignty in the Age of Assisted Decision-Making

Generative AI systems now mediate everyday decisions in healthcare, education, law, and work, at a scale we have never seen before. The dominant response to this phenomenon frames it as a challenge of individual digital literacy and risk-based regulation. We argue that this framing is structurally insufficient for emerging economies, because it presupposes conditions of equity that simply do not exist: the infrastructure processing these decisions is concentrated in foreign jurisdictions, the training data systematically underrepresents peripheral populations, and the regulatory frameworks are defined by actors these economies do not control. To address this structural gap, we propose the concept of structural cognitive sovereignty: the collective and institutional conditions necessary for genuine cognitive autonomy to be possible in the era of assisted decision making. The concept articulates three constitutive dimensions, infrastructure control, data representativeness, and regulatory autonomy, and argues that theseare necessary conditions, not optional aspirations. Drawing on the philosophy of cognitive autonomy in Aristotle, Kant, and Sartre, as well as token economics, neurosymbolic AI architecture, and speculative design, we document how the same AI subscription represents 7.8 times the income burden for a Brazilian professional as for an American counterpart, and how tokenization inefficiency imposes an additional surcharge of 31 to 63 percent on Portuguese-language users. Two speculative scenarios for Brazil in 2035 illustrate trajectories of structural dependence versus structural sovereignty, grounded in existing precedents from the Global South including the Masakhane initiative and India’s Bharat-Gen program. A research agenda is proposed for emerging economies seeking to move from cognitive dependence to structural sovereignty.