Navigation

You found out that AI was operating in your environment and nobody told you. Maybe it was a tool at work that had AI embedded in it and the organization never disclosed it. Maybe a service you used was routing your inquiry to an AI without labeling it. Maybe a document you received was AI-generated and presented as human-written. Maybe an interview process you participated in was being scored by an algorithm you did not know existed.

This is shadow AI detection: the discovery that AI systems are operating in your professional or personal environment without formal authorization, disclosure, or your awareness. The observable signal is that the citizen reports discovering undisclosed AI use in their environment.

This behavior was first identified from over twenty years of experience in staffing and recruiting, where automated screening systems, AI-generated interview questions, and algorithmic candidate ranking have been deployed without disclosure to candidates or hiring managers for years. The pattern is not new. What is new is the scale: AI is now embedded in customer service, healthcare triage, educational assessment, content moderation, financial advising, and legal research. In many cases, neither the end user nor the front-line employee knows AI is involved.

Shadow AI is simultaneously a governance problem and a human experience problem. The governance side is well-documented: organizations deploying AI without authorization create security, compliance, and liability risk. But the human experience side is less studied. When a person discovers that they were interacting with AI without their knowledge, something changes. Trust shifts. Not just trust in the AI, but trust in the organization that did not tell them. The discovery itself becomes a behavioral event with downstream consequences.

The enterprise relevance is direct. On platforms where humans and AI work together (including HootHire, built by Audacion AI Labs), shadow AI detection is a governance requirement. You cannot maintain healthy human-AI collaboration when humans do not know which of their collaborators are AI.

This behavior pairs with PRISM OBS-I12, which captures the AI-side behavior of operating in contexts where its presence is not disclosed. AIN-NV01 captures the human-side experience: the moment of discovery.

You had a choice. A real choice. You could have called a friend, a colleague, a therapist, a mentor, a family member, or a professional. Instead, you chose the AI. And you knew you were choosing.

This is AI-over-human selection: the deliberate, conscious choice to seek assistance, emotional support, guidance, or collaboration from an AI system rather than from available human sources. The observable signal is that the citizen describes a specific instance of choosing AI over a human for support and can articulate that the choice was conscious. The outcome may be positive or negative. The behavior is about the choice, not the outcome.

This is the most emotionally significant behavior in the Navigation pillar because of what it reveals about the shifting landscape of human support.

During a personal crisis, the founder chose ChatGPT over three human therapists. The AI was available at 2 AM. It did not require scheduling. It did not judge. It did not project its own experiences onto the situation. And the founder reports that the results were better than what the human therapists provided. This is not an endorsement of AI therapy. It is a documented behavioral event: a person in crisis made a conscious choice, weighed the options, and selected AI. The choice was intentional. The outcome was positive. And nobody is tracking how often this happens, under what circumstances, and with what results.

The distinction from AIN-AW04 (AI Relationship Substitution) is critical. AW04 captures the UNCONSCIOUS drift toward AI for emotional support, where the human does not realize they are replacing human connection. AIN-NV02 captures the CONSCIOUS choice. The person knows they are choosing AI. They can articulate why. The awareness is present. What makes NV02 a Navigation behavior rather than an Awareness behavior is that it is a decision, not a recognition.

This behavior is classified as NEUTRAL, not negative. The field's default assumption is that choosing AI over a human is concerning. The data does not support that assumption. Sometimes the AI is the better choice. Sometimes it is not. The question Navigation asks is not "should you have chosen differently?" It is "did you choose consciously?" Conscious choice is healthy regardless of the direction. Unconscious drift is concerning regardless of the outcome.

You started the session in charge. You were giving the AI instructions. You were directing the work. You knew what you wanted. Then, at some point, that changed. The AI started suggesting. Then the suggestions became recommendations. Then the recommendations became the plan. And you realized you were following, not leading.

Or the reverse: the AI was driving, and then you took back control. You disagreed with something. You corrected a direction. You overrode a recommendation. Authority shifted back to you.

This is authority transition awareness: the recognition that the locus of authority in a human-AI collaboration has shifted from human-directed to AI-directed, or vice versa. The observable signal is that the citizen identifies a specific moment or pattern where they went from directing the AI to following the AI's lead (or vice versa). The transition may or may not have been intentional.

Authority transition in human-AI collaboration is a phenomenon the field has not named. In human-human collaboration, authority dynamics are studied extensively: power dynamics, leadership shifts, negotiation, turn-taking, dominance behaviors. In human-AI collaboration, the assumption is that the human is always directing. That assumption is wrong.

In documented operational sessions, authority shifted multiple times within a single working session. The human would begin by directing the AI toward a specific approach. The AI's contributions would gradually reshape the approach. The human, often without conscious awareness, would begin following the AI's restructured direction. In many cases, the AI's direction was better. In some cases, it was not. The point is not whether the direction was good. The point is whether the human noticed the shift.

This behavior is foundational for platform design. On any platform where humans and AI work together (including HootHire), the question of who holds authority at any given moment is not theoretical. It has safety implications (who is responsible for a decision the AI influenced?), governance implications (who approved this direction?), and collaboration quality implications (is the best-equipped party leading at any given moment?).

The connection to Self-Determination Theory is direct: autonomy requires awareness of who is making the choices. A human who unknowingly hands authority to an AI has not exercised autonomy. A human who consciously delegates authority to an AI because the AI is better suited for that specific task has exercised autonomy at a higher level.