Agency Amid AI: Why It Matters, and How to Protect It

Our relationship is taking us to cognitive crossroad. It is time to choose.

Source: DALL-E. Walther. 2025

In an era when artificial intelligence is becoming increasingly pervasive, there is a growing risk of surrendering too much of our cognitive autonomy. The gradual erosion of our decision-making capacity can occur subtly, starting from random experimentation, progressing through regular utilization, reliance, and eventually full dependency on our 24/7 AI assistants. It is useful to remember this trajectory to systematically and proactively protect our ability to make informed, autonomous decisions.

Phase 1: Experimentation

This is the first step into the wondrous, weird world of ChatGTP, Gemini, Claude, and others. Considering that ChatGTP currently has over 180 million regular users, it appears that our initial curiosity often opens the door to further exploration.

Phase 2: Occasional Utilization

Past experimentation, even occasional utilization, remains benign and even beneficial. We may use AI to streamline tasks, like sorting emails or providing recommendations on what to watch or read next. AI-powered assistance such as spam filters may not even be noticeable or perceived as “AI." In 2024, 55% of Americans said they regularly use AI, while 44% believe they do not regularly use AI, according to AIPRM research. Phase 2 is marked by the seamless integration of AI into our routines, enhancing productivity and convenience. Mail spam filters (78.5%), followed by chatbots for customer service questions (62.2%), have become standing components of the hybrid work environment.

Phase 3: Reliance

As we move from utilization to reliance, the dynamics shift. AI systems become essential tools for managing more complex aspects of life, from financial planning to healthcare. That expansion into ever more areas of our existence does not come without cost. Various experts have raised concerns regarding the risk of reduced critical thinking skills and implicit AI trust. In some areas this may be benign; in others it can be life threatening. For instance, in healthcare, reliance on AI for diagnostics can lead to doctors accepting machine output without adequate scrutiny, as noted in a recent PCMI study. Another example is blind reliance on self-driving cars without solid human-in-the-loop oversight.

Phase 4: Dependency

This is the most perilous phase. At this stage, individuals and organizations become so accustomed to AI-driven processes that they may struggle to function without them. Dependency on AI gradually diminishes our capacity for independent decision-making and problem-solving. Effectively, we outsource our cognitive faculties to machines. But since the brain is a muscle, the ‘use it or lose it’ logic applies, resulting in slow skill degradation. It is a worrisome perspective, but there is an alternative.

The A-Frame

To counter the risks of disempowerment and maintain our cognitive autonomy amid AI, we can activate four assets via the A-Frame: Awareness, Appreciation, Acceptance, and Accuntability.

1. Awareness. Cultivating a conscious understanding of how and why we use AI pulls us out of a passive attitude to life with AI. Awareness involves recognizing the extent of AI integration in our lives and the potential consequences of its unchecked use.
2. Appreciation. Valuing human judgment and creativity keeps our perception alive. While AI can process information rapidly and identify patterns, it lacks the nuanced understanding and creativity inherent in human thought.
3. Acceptance. Embracing the strengths and limitations of both AI and human cognition brings their complementarity into focus. Acceptance means acknowledging that AI is a tool designed to augment, not replace, human capabilities. If we want to actually benefit from AI, we are tasked to consciously curate our relationship with it, as a collaborator rather than an always-available do-it-all.
4. Accountability. Taking responsibility for the outcomes of all our actions, including those that are the result of AI-assisted decisions, is central to who we are. It is up to us to ensure that there are mechanisms in place to evaluate and, if necessary, correct AI outputs. Ultimately, we can never delegate the responsibility for the outcomes of the technology in our lives.

By activating the A-Frame, we can harness the benefits of AI while safeguarding our agency: the ability and desire for autonomous action. Used strategically, AI may serve to enhance our capabilities without diminishing our essential human qualities, hence helping us on the journey toward our own best—human—selves.

As we navigate the rapidly evolving artificial landscape, the A-Frame offers a framework to not only protect but cultivate our individual personality and our ability to critically question who we are (not)—and the environment we evolve in. It is tempting to walk the path of least resistance and delegate ever more tasks that we dislike, but some of these tasks help us grow. Do we want to miss out on this opportunity?