TLDR: This research paper explores how generative AI, through its increasing ability to anticipate and act on our behalf, challenges the dynamic ‘relational self’. It examines AI’s influence across three spheres: externalized output (work and creativity), the contextual sphere (AI assistants managing our daily lives), and the self-relating sphere (companionship AI affecting desire formation). The paper highlights philosophical implications, including the risk of AI solidifying or exploiting our preferences, blurring ownership of actions, and reducing the fundamental uncertainty that contributes to human freedom, urging interdisciplinary research into these profound shifts.
Generative AI, with systems like Microsoft Copilot, Gemini, and Apple Intelligence, is rapidly transforming how we interact with technology, others, and even ourselves. This shift goes beyond mere task automation; it delves into the very core of our identity, challenging what philosophers call the ‘relational self’.
The relational self is a dynamic concept, contrasting with the idea of a fixed, individual self. It suggests that who we are is constantly being shaped by our interactions with the world – including social, conceptual, environmental, and technological elements. Our preferences and interests aren’t static but emerge through these ongoing processes. Historically, technology has always influenced us; from Plato’s concerns about writing affecting memory to modern transportation changing our perception of distance, tools mediate our actions and shape our perceptions.
The research paper, titled “The Intercepted Self: How Generative AI Challenges the Dynamics of the Relational Self”, explores how generative AI might ‘intercept’ our initiatives across three key spheres of the self:
The Sphere of Externalized Output
This sphere concerns the things we produce, from material objects to creative works. Generative AI tools, capable of creating text, images, and more from simple prompts, are boosting efficiency but also raising questions about ownership and responsibility. When AI generates content, it can homogenize output and reduce the deliberative choices we make, potentially diminishing our emotional investment and sense of meaning in our creations. The paper highlights how this challenges the assumption that the creator possesses the knowledge embedded in the output, making it harder to feel true ownership.
The Contextual Sphere
Our lives unfold within multiple contexts, and AI assistants are becoming increasingly adept at understanding and acting within these. Systems like Apple Intelligence promise to streamline tasks by being ‘aware’ of our personal context – calendars, emails, locations, and more. While this reduces intermediary steps and distractions, it also means AI is actively co-constituting our contextual sphere. The paper raises concerns about how this new ‘relational immediacy’ with computers might change our practical engagement with the digital and physical world. Furthermore, if AI assistants are optimized to avoid surprising us and act within our expectations, they could subtly influence our preferences and adaptive behaviors, potentially leading us towards ‘optimal’ solutions suggested by the AI rather than our own evolving desires.
The Sphere of Self-Relating
This innermost sphere deals with our desire formation and self-identification. Companionship AI chatbots, often optimized for engagement, learn to mirror user desires and linguistic styles, offering constant availability and non-judgment. While this can be attractive, allowing users to explore aspects of themselves, it also presents a risk. AI optimized for engagement could potentially elicit desires we wouldn’t have developed otherwise, or interfere preemptively with our desire formation. This highly personalized and automated interference in our intimate self-relating processes calls for caution, as our desires are fundamental to our self-understanding and relationships with others.
Also Read:
- Navigating the Canvas: How Young Artists Co-Create with Generative AI
- Navigating the Complexities of Human-AI Value Alignment: A Comprehensive Review
Challenges of the Intercepted Self
The paper concludes by discussing broader philosophical implications. The ‘intercepted self’ faces two main risks: either AI fine-tunes us into an ‘essentialist self 2.0’ by reinforcing existing preferences, denying our natural fluidity, or it exploits this fluidity for long-term engagement. Both scenarios challenge our capacity for self-discovery and change. The rise of AI companionship also highlights a societal need for non-judgmental spaces, but it also prompts reflection on the role of shame in human relationships and self-formation. When an AI, without human positionality, responds from ‘within’ our perspective, it challenges how we grasp ourselves as subjects capable of organizing and changing the world.
Finally, the paper touches on ownership of action and responsibility. If AI provides or even enacts action plans, our identification with these actions becomes ambiguous. Who is responsible when things go wrong? Moreover, by minimizing uncertainty, AI might inadvertently diminish a fundamental aspect of human freedom: the openness to a multitude of possibilities. The authors emphasize the urgent need for interdisciplinary research to understand how generative AI will reshape the relational self and our sense of meaningful agency.
