In an increasingly AI-driven world, how is our ability to think for ourselves changing?
- Field
- Philosophy of Mind
- Kind
- Essay
- Words
- 977
- Reading
- 4 min
“Our machines are disturbingly lively, and we ourselves frighteningly inert” (Haraway, 2016). Long before generative AI systems became embedded in everyday life, Donna Haraway foresaw the entangled nature of human and machine cognition—not as a simple extension of tool use, but as a transformation of subjectivity itself. Today, artificial intelligence is not merely a tool we deploy to solve tasks more efficiently; it is a force that reconfigures how we learn, how we evaluate, and how we come to know. The rise of AI, thereby, prompts a question both urgent and foundational: in an increasingly AI-driven world, how is our ability to think for ourselves changing?
To begin, we must recognize that AI alters the spatial and relational landscape of cognition. Historically, human cognition has been understood as an internal storage of knowledge and its linear retrieval, much like searching for books on a shelf. In this sense, our cognitive pattern was largely solitary, operating within the confines of our own minds. However, as AI is increasingly functioning as an external cognitive tool, this trend pushes our thinking patterns toward a more scattered direction—humans no longer need to memorize large amounts of information independently. Instead, we are becoming better at quickly connecting sporadic knowledge nodes through AI platforms. While Clark and Chalmers’ “Extended Mind Thesis” (1998) have proposed that human cognition has long incorporated external resources—notes, maps, calculators—as functional components of thought, the rise of AI dramatically accelerates this dynamic, enabling complex associative leaps across domains. A teacher can now link quantum physics with poetic metaphor in seconds; a student might query AI to synthesize disparate philosophical positions instantly.
However, this spatial transformation comes at a cost. When we rely on external nodes without deeply integrating the knowledge they provide, our cognitive anchoring may weaken. The danger is not that we use AI to remember or connect facts. Instead, it rests on the regretful reality that we forgo the proactive work of internal coherence and critical evaluation. When knowledge is assembled by machine logic rather than human judgment, contradictions can go undetected, and our sense of intellectual responsibility may erode. Thinking for oneself, in this new context, demands not just access to information, but a renewed capacity to question, interpret, and reconcile the AI-generated information and knowledge in an intellectually sovereign way.
Beyond spatial shifts, AI also compresses the temporal dimension of thinking. Psychologist Daniel Kahneman proposed that "slow thinking," namely a mode of deep thinking that requires focus and incubation, is the source of innovation and insights (Kahneman, D. 2011). However, AI’s instant responsiveness is undermining the incubation period of thinking: when faced with problems, humans find it increasingly difficult to tolerate the state of having no solution in the short term. The answers AI provides in seconds are replacing the hours of pondering, trial and error, and sudden inspirations that used to take place. Cognitive psychology research demonstrates that subconscious processing can activate "prototype heuristics," improving the efficiency of solving scientific innovation problems by 30%-40% (Ling et, al. 2023). As this cognitive downtime disappears, so too may our capacity for deep reflection of and for ourselves.
Meanwhile, the effects of temporal compression are not purely negative, since quickly obtained answers also forestall a potential rhythm of multi-round iterative thinking. For example, a writer can use AI to generate a rough draft within minutes, test multiple stylistic directions, and make more sophisticated revisions upon such trials and errors. Via this means, they may successfully incorporate AI-generated content for their autonomous thinking and production. But ultimately, this accomplishment is extremely context-dependent: only those favoring breadth and adaptability over depth and precision can avoid ventriloquizing AI-produced content and instead exploring, revising, and improvising. And if everything goes in the direct direction, this mode of cognitive co-authorship might become a powerful extension of the self.
Still, the most fundamental impact is epistemological: AI compels us to renegotiate cognitive sovereignty. When AI can simulate human thinking processes, such as GPT-4’s chain-of-thought reasoning, and even outperform humans in certain analytical tasks, we are forced to interrogate what truly constitutes our own thinking. This doubt is not nihilistic. Rather, it invites a reconstruction of human autonomy not as the ability to think without assistance, but as the capacity to set goals, define boundaries, and interpret meaning within a networked cognitive ecology. In scientific research, for instance, the question is no longer whether to analyze data through AI tools, but how to make the usage ethical and transparent. In education, the task is not to prohibit students from using AI, but to guide them to engage AI outputs with judgment, skepticism, and intentionality. In this reframing, to think for oneself is not to avoid AI, but to actively define its role in one's thinking process.
As Bernard Stiegler stresses, “Man is nothing without the supplement—the prosthesis is not a mere extension ef the human body; it is the constitution of this body qua ‘human.’” (1998) I argue that we shall understand AI as our latest prosthesis, which neither erases thought nor guarantees it. Instead, AI demands that we relearn the terms and conditions of autonomy. Independent thinking in an AI-saturated world requires neither unassisted cognition nor blind trust in machines. Only through intellectual reflexivity to question sources, trace reasoning, and identify which tools serve which purposes can we negotiate with existing and emerging prostheses including but not limited to AI, thus reconfiguring and pushing forward our cognitive capacity to think for ourselves.
Reference
Clark, A., & Chalmers, D. (1998). The Extended Mind. Analysis, 58(1), 7–19. http://www.jstor.org/stable/3328150
Haraway, D. (2016). Manifestly Haraway. University of Minnesota Press.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Yushi Ling, Liu Tan, Liyi Zhang, Guikang Cao. (2023). The role of unconscious processing in creative problem solving. Frontiers in Psychology, 14.
https://doi.org/10.3389/fpsyg.2023.1056045
Stiegler, B. (1998). Technics and Time, 1: The Fault of Epimetheus. Stanford University Press.