There is a noticeable shift in how we speak about artificial intelligence. Not long ago, the emphasis was on generation. Systems produced text, images, and code, often with surprising fluency. They responded to prompts, extended ideas, and supported human creativity. The relationship was clear. We asked, and they answered.
That clarity has started to blur. The conversation now turns toward systems that act, not only in response to instructions, but in pursuit of goals. These systems can plan, call tools, adapt to changing inputs, and carry tasks forward across multiple steps. Increasingly, they are described using a more loaded term. They are said to “decide.”
This shift in language carries more weight than it first appears. To describe something as making decisions is not a neutral technical claim. It draws from a long history of thinking about agency, intention, and responsibility. Even in everyday use, the word suggests ownership. A decision is not just an output. It implies a moment of commitment, a selection among possibilities that carries meaning.
Yet when we apply this word to machines, something feels unsettled. We recognize that these systems are built, trained, and constrained. Their behavior is shaped by data, architecture, and design choices. At the same time, their actions can appear coherent, adaptive, and at times even purposeful. The tension is not easily resolved.
Rather than rushing to define what machines are doing, this moment invites a different kind of attention. If we are willing to pause, the question begins to turn back toward us. What have we always meant when we say that we decide?
The Familiar Story of Free Will
In ordinary life, decision making feels straightforward. We imagine ourselves as agents who consider options, weigh outcomes, and choose accordingly. The process appears centered. There is a sense of an “I” that evaluates and selects, even when the choice is difficult or uncertain.
This experience carries a strong sense of ownership. When we say “I decided,” we are not only describing an action. We are affirming a relationship between ourselves and what follows. The decision becomes part of our identity. It reflects our values, our judgment, and our responsibility.
At the same time, a closer look reveals that this picture is more layered than it appears. Many influences shape our choices before we become fully aware of them. Habits formed over years guide our reactions. Emotions arise quickly, often before deliberate reasoning begins. Social expectations, cultural norms, and situational pressures all play their part.
Even when we believe we are reasoning carefully, our thinking is not detached from these conditions. What feels like a clear evaluation may already be directed by prior inclinations. The set of options we consider is often limited by context. The criteria we use are shaped by what we have learned to value.
None of this eliminates the experience of deciding. The sense of choice remains real. Yet it becomes harder to locate a single point where the decision originates. The process unfolds across layers, some visible, some not. The idea of a fully independent act of will begins to feel less certain.
Philosophical traditions have long explored this tension. Some have argued that free will is an illusion, a story we tell after the fact. Others have defended it as a meaningful feature of human life, even within constraints. What remains consistent is the difficulty of defining it in a way that fully matches our experience.
Systems That Reflect Our Structure
When we turn back to contemporary AI systems, the resemblance becomes more striking. These systems are built from components that process input, store information, evaluate options, and produce outputs. When arranged in sequences and loops, they can carry out extended tasks that appear organized and goal-directed.
A system that plans a series of steps, adjusts based on feedback, and selects among alternatives begins to resemble a decision maker. It does not simply react. It operates across time, integrating past information with current conditions and anticipated outcomes.
What makes this especially interesting is not that machines have suddenly acquired human qualities. It is that they make visible a structure that may already be present in our own decision making. The layers that we tend to compress into a single act are here separated and observable.
We can see how inputs are filtered, how options are generated, how criteria are applied, and how selections are made. The process is explicit, even if the internal mechanics remain complex. What appears in machines is not identical to human cognition, but it is close enough to invite comparison.
This comparison does not reduce human experience to computation. It does, however, challenge the assumption that decision making is a simple, unified act. The presence of similar patterns in artificial systems suggests that what we call deciding may be less singular than we often assume.
Rather than asking whether machines truly decide, it may be more productive to ask what kind of process gives rise to the appearance of decision in both humans and systems. The answer seems to involve layers, feedback, and the capacity to evaluate possibilities over time.
The Dissolving Center of the Self
If decision making is layered, then the idea of a central decision maker becomes harder to sustain. The image of a self that stands apart from its conditions and chooses freely begins to shift. Instead of a single point of origin, we find a field of interacting processes.
In this view, the self does not disappear, but its role changes. It becomes less a source and more an organizer. It gathers experiences, interprets them, and forms a coherent narrative. It provides continuity across time, allowing us to understand our actions as part of a larger story.
Narrative plays a crucial role here. After a decision is made, we explain it. We describe our reasons, our intentions, and our goals. This explanation creates clarity. It allows us to take responsibility and to communicate with others. It also simplifies what may have been a complex process.
The narrative gives the impression that the decision originated from a unified center. It presents a clear line from intention to action. Yet this clarity is, in part, constructed. It is shaped by the need for coherence, not only by the structure of the process itself.
This does not make the narrative false. It makes it selective. It highlights certain elements while leaving others in the background. The result is a stable sense of self that can act, reflect, and relate to others.
In light of this, the question of free will becomes less about absolute independence and more about how we participate in these processes. The self is not outside the system. It is one of its most important features.
Delegating the Decision Loop
As AI systems become more capable, we begin to delegate parts of this layered process. Tasks that once required human attention are now handled by systems that can evaluate, prioritize, and act within defined boundaries.
This delegation is not uniform. In some cases, systems provide recommendations that humans review and approve. In others, they carry out actions directly, with human oversight occurring afterward. The structure of decision making is being rearranged.
The change is especially visible in environments that require rapid response. In cybersecurity, for example, systems can correlate signals, identify patterns, and suggest actions in real time. Contemporary AI-driven systems show how detection, analysis, and response can merge into a continuous, self-adjusting process.
In such contexts, the question is not whether the system decides in a human sense. It is whether it can participate effectively in the decision loop. Can it identify what matters? Can it act within constraints? Can it adapt as conditions change?
As these systems become more reliable, the balance shifts. Humans move from direct control toward supervision. The role changes from making each decision to shaping the conditions under which decisions are made.
This shift raises important questions. How do we assign responsibility when actions are distributed across human and system components? How do we maintain trust when parts of the process are no longer directly visible? What does it mean to exercise judgment in a system that already evaluates and acts?
These questions do not have simple answers. They point to a deeper transformation in how action is organized.
Not Free, Not Mechanical
At this point, the familiar binary between free will and mechanism begins to lose its clarity. The evidence from both human experience and artificial systems suggests that decision making does not fit neatly into either category.
If we insist on complete freedom, we struggle to account for the many influences that shape our choices. If we reduce everything to mechanism, we lose sight of meaning, interpretation, and responsibility.
A more balanced view recognizes that decision making is structured and constrained, yet still meaningful. It unfolds within conditions, but it is not exhausted by them. There is room for interpretation, for reflection, and for change.
In this sense, human agency can be understood as participation in a process rather than control from outside it. We do not stand apart from the forces that shape us. We engage with them, sometimes consciously, sometimes not.
Agentic AI does not resolve this tension. It brings it into clearer view. By building systems that mirror aspects of our own processes, we are encouraged to reconsider what we have taken for granted.
The presence of these systems does not diminish human agency. It reframes it. It shows that agency has always been more layered, more relational, and more dependent on context than a simple model of free will would suggest.
Living Within the Question
What remains is not a final answer, but a shift in how we approach the question. Decision making can no longer be treated as a simple act of will. It is better understood as a dynamic process that involves multiple layers of influence, interpretation, and action.
This understanding does not make action meaningless. It invites a different kind of attention. If our decisions arise within conditions, then awareness of those conditions becomes part of what it means to act well.
In a world where systems participate in decision processes, this awareness extends beyond the individual. It includes the structures we design, the tools we use, and the relationships we form with them. Agency becomes something that is shared, not in the sense of being diluted, but in the sense of being distributed.
To live within this perspective is to accept a certain complexity. We continue to make choices, to take responsibility, and to seek clarity. At the same time, we recognize that the ground beneath these actions is not as simple as it once seemed.
The emergence of agentic AI does not close the question of free will. It keeps it open in a new way. By placing decision making in a broader context, it allows us to see both its structure and its significance with greater precision.
What changes is not that we stop deciding. It is that we begin to see decision not as a singular moment, but as an ongoing process in which we are always already involved.
And in that recognition, the meaning of action remains, not as absolute freedom, and not as mere reaction, but as something that takes shape within the conditions we inhabit and the understanding we continue to deepen.
Image: StockCake
Tom’s Blog 