From Machado de Assis to Cybersecurity: Not that kind of AI Literacy Course

Fernanda Eliott (written by Eliott only, no GenAI), Spring 2026.

  • How to cite this artifact: Eliott, F. (2026, July 8). “From Machado de Assis to Cybersecurity: Not That Kind of AI Literacy Course.” Cognitive Robustness Studio: Conceptual Precursors. https://doi.org/10.17605/OSF.IO/P7D3Q

   The cognitive degradation path technocentrism has been paving evokes the sense that we deserve a satirical critique of science, institutions, and power, as Machado did in The Alienist. This work arose from such a feeling and a thought experiment that breaks into two:

  1. Human-AI Cognitive Signature Spill: What if someone’s way of thinking (cognitive patterns, including affective residues) spills over into Generative Artificial Intelligence (GenAI) tools, through repeated interaction and model adaptation, without user awareness or consent? What if, as GenAI tools proliferate globally, they are not just exporting capabilities, but importing cognition, performing epistemic extractivism? (What if human mosaics are vacuumed by these tools?)
  2. And what if that helps to build a composite artificial intelligence, later sold back to us (potentially as a sort of distributed AGI)? Isn’t it convenient for industry to label a worker’s intelligence in terms of Agentic Work Units (AWUs) and, concomitantly, turn intelligence into a commodity that people can buy from a machine?

   Those questions would have sounded bizarre five years ago. Aiming to equip undergraduate students to face today’s civilizational shift, this work provides a critical evaluation of a transdisciplinary, summer-long intervention that can be described from two angles: core computer science (focusing on Reinforcement Learning and Multi-Agent systems) and humanistic. Here, I focus on the humanistic. The intervention serves as an experimental pedagogical lab, using zines, DIY culture, and philosophy, as well as Brazilian literature and art (e.g., Machado de Assis, José de Alencar, Oswald de Andrade, and Tarsila do Amaral’s Abaporu). The pedagogical choices challenge tech’s hegemonic constructs. More than that, the intervention explores linkages between the Skinner box and GenAI and covers twelve umbrella topics (each with readings and activities), ranging from the Importance of Experiencing Discomfort, AI Hygiene, and Persona Parasitology to Social Engineering, GenAI Benchmarks, Cybersecurity, and Red Teaming.

   The intervention asks students to pursue three overarching reflections: 1) What is a Resilient Mind? What is cognitive robustness? 2) What is and how to Untrain Predictability? 3) What should an AI literacy course cover? What is AI hygiene? In this work, I reflect on the intervention’s pedagogical choices, outline qualitative metrics, and assessments (an IRB was obtained to conduct the intervention). Finally, as it is my claim that an AI Literacy course should be transdisciplinary and cover much more than technical skills, the intervention provides pathways for designing Not that kind of tech-submissive AI Literacy Course.

Notes:

  1. This abstract was submitted in Spring 2026 and later became a conceptual precursor to the Cognitive Robustness Research Studio. It is preserved here as part of the Studio’s intellectual genealogy.
  2. Hyperlinks added here for better contextualization. 
  3. The term “persona parasitology” is inspired by The Rise of Parasitic AI, by Adele Lopez.