What Surprised or Unsettled Me

Of all the ideas presented across both materials, the one that unsettled me most was Dr. Zaki’s argument that AI companionship may gradually “erode our capacity to tolerate the difficulty of real human relationships”. Her point was not simply that AI is addictive or convenient; it was something more structurally concerning. That by acclimatizing us to responses that are immediate, perfectly attuned, and never burdensome, AI may quietly set different expectations of what we want from people; and by nature, people will always fall short of that standard. They are slow, inconsistent, distracted, and sometimes wrong. Those imperfections are not flaws in human connection, they are the substance of it. What worried me about Dr. Zaki’s thesis is that I am already watching it happen. In university settings around me, students increasingly default to AI tools not just for academic tasks but for the kind of processing that used to require and strengthen human relationships such as working through confusion, seeking reassurance, and thinking out loud. The option to avoid the ‘friction’ of asking another person, of being vulnerable or potentially judged, is now always available. And I see people taking it, not out of laziness, but because AI offers something real relationships cannot always guarantee; unconditional availability and speed. That is a deeply human need being met by something that cannot truly meet it.

The second thing that surprised me came from Dr. Gebru’s podcast; the scale and invisibility of the human labor that sits behind AI systems. I use AI tools daily for engineering tasks, for structuring work, or streamlining processes and I had genuinely not considered the full chain of human cost behind them. The workers are paid almost nothing to annotate data, to moderate toxic content, to absorb the psychological damage of what these systems process. Dr. Gebru’s point that this phenomena is not accidental, but structural, to help big data companies be profitable and look good on the books. This was not something shocking for me as all companies want to be profitable, but it was still an uncomfortable eye-opener. For instance, the “data cloud” narrative exists partly to obscure the very human infrastructure beneath it. For someone who thinks of themselves as ethically aware, realizing I had been using these tools without interrogating their foundations was a genuine moment of accountability.

Connection to the Course

Dr. Gebru’s analysis of AI as a mechanism for concentrating power among a small number of Silicon Valley corporations mirrors the same logic we examined in the Voter Suppression activity and in Mounzer’s writing: systems that appear neutral or even beneficial on the surface, while quietly reproducing inequality at a structural level. The workers exploited to train these systems are disproportionately from the Global South; a detail that connects directly to the course’s discussion of othering and whose labor, suffering, and dignity is made invisible in the dominant narrative. Dr. Zaki’s work connects equally to the course’s emphasis on empathy and human connection. Her warning that AI simulates resonation and compassion without providing it echoes the depression simulation we completed which showed that genuine support requires a human capacity to sit with discomfort, not just to respond to it. Both materials, taken together, suggest that the forces this course has asked us to resist dehumanization, the erosion of equity, and the flattening of individual experience are not only political or historical. They are also technological, and they are accelerating.

What I Will Do Differently

These materials have sharpened a principle I was already beginning to practice but had not fully adopted yet. I aim to start using AI only where it genuinely frees up cognitive capacity for work that requires human judgment, creativity, or connection and not as a substitute for the effort that builds those capacities in the first place. As an engineer, there are genuinely redundant processes where AI streamlines work without replacing thinking. That use I can defend without being penalized or needing additional human capacity. What I am more deliberate about now is the boundary; the moment AI starts substituting for the discomfort of real intellectual or emotional engagement, it is no longer a tool, it is an avoidance mechanism. Now knowing the human cost behind these systems, such as the underpaid labor and the psychological damage absorbed by content moderators, also means I now carry a responsibility to use them with that awareness rather than letting the convenience of the interface make the cost invisible again.

What the Course Should Cover in Future

The gap I noticed most in the course is the specific psychological impact of AI on attachment and human connection within Arab societies. Dr. Zaki’s lecture is a valuable starting point, but the course would benefit from exploring how loneliness, digital dependency, and the erosion of traditional social structures intersect in contexts like Egypt, where collective identity and family-based support systems are culturally central. When those structures are disrupted, whether by migration, economic pressure, or rapid digitization, the appeal of AI companionship may be especially powerful, and the risks Dr. Zaki describes may be especially acute.