Integration

Your work process now structurally depends on AI. Not emotionally (that is AIN-IN03, Reliance Dependency in the Independence pillar). Structurally. You have built workflows, templates, habits, and sequences that include AI as a load-bearing component. When AI is unavailable, the process itself degrades. Tasks take longer. Output quality drops. Some tasks cannot be completed at all because the workflow was designed with AI as part of the infrastructure.

This is AI workflow dependency: structural dependency of the human's workflow on AI availability, where the removal of AI produces measurable degradation in process efficiency, output quality, or task completion capability. The observable signal is the citizen reporting that specific work processes degrade or fail without AI.

The distinction from AIN-IN03 (AI Reliance Dependency) is precise and important. IN03 is about the human: you feel anxious, lost, or unable to function. IG01 is about the process: the workflow itself breaks, regardless of how you feel about it. A calm, emotionally resilient human can still have a structurally dependent workflow. The human is fine. The process is stuck.

In documented operational research, workflow dependency appeared most clearly in multi-step creative and analytical processes where AI handled intermediate steps. A human who used AI for research synthesis, drafting, and structural review had built a three-stage workflow with AI embedded in all three stages. When AI was unavailable, each stage required manual execution at significantly lower speed. The workflow was not merely slower. It was architecturally different without AI, requiring the human to perform tasks they had stopped building the infrastructure to handle independently.

You had a question, a problem, a task. And your first move was to open AI. Not because the task was complex. Not because you needed specialized knowledge. Because AI is where you start now. It is your default. The habit formed gradually: AI gave good answers, so you went there first. Then you went there first every time. Then you stopped considering the alternative. The sequence "think about it, then ask AI" became "ask AI."

This is AI-first defaulting: habitual initiation of AI assistance as the default first step in problem-solving, analysis, or creative tasks, bypassing the human's own cognitive engagement with the problem. The observable signal is the citizen reporting that they open AI before attempting independent thought. The pattern is habitual, not occasional.

This behavior was pressure-tested for pillar placement. It could be an Independence behavior (you are losing the habit of independent thought). It is classified as Integration because the core observation is a pattern change: how you start work has structurally changed. AI-First Defaulting is about your workflow default, not your cognitive capability. You may still be perfectly capable of independent thought. You have simply stopped starting there.

The distinction matters for intervention. If this were an Independence problem, the intervention would be cognitive: rebuild the thinking muscle. Because it is an Integration problem, the intervention is structural: redesign the workflow so that independent thought precedes AI consultation. Different diagnosis, different prescription.

In documented operational research, AI-First Defaulting appeared as a gradual habit consolidation. The human initially used AI for complex tasks, then for moderate tasks, then for tasks they could easily have handled alone. The shift from "AI for hard things" to "AI for everything" happened without a decision point. It accumulated through repetition until the default was no longer a choice but a reflex.

AI made you feel like you could do more. So you did more. You took on broader scope. You said yes to projects you would have declined a year ago. You worked later because the work was flowing and AI was making it flow faster. You expanded into domains you had never operated in because AI made it feel accessible. The hours went up. The scope went up. The intensity went up. And it felt great. It felt like empowerment. It felt like capability.

The data tells a different story.

This is AI-enhanced work intensification: measurable increase in work hours, scope, pace, or intensity attributed to AI collaboration, often accompanied by a subjective perception of enhanced productivity that may not correspond to objective output measures. The observable signal is the citizen reporting increased work hours, expanded scope, or accelerated pace. The key indicator is the perception-reality gap: believing productivity increased when it may not have.

The Berkeley Haas study (2026) is the primary evidence. Over eight months, 200 employees at a technology company showed an average increase of 3 hours and 15 minutes of additional work per week after AI integration. Workers described the experience as positive: they felt more productive, more capable, more engaged. The researchers identified this as work intensification, the same pattern Francis Green (2004) documented across every wave of workplace technology since the Industrial Revolution. The pattern is consistent: new technology promises efficiency, delivers intensification, and the feeling of empowerment masks the structural acceleration.

The METR study compounds the concern. Developers believed AI made them 20% faster. The data showed they were 19% slower. A 39-percentage-point perception gap. The DORA 2025 report confirmed the pattern: developers using AI worked more hours while quality metrics did not reflect the perceived productivity gains. The ActivTrak report (2026) documented the same trend across 1,111 organizations and 443 million hours of work activity.

AIN-IG03 is one of the most well-documented human behavioral changes in AI collaboration. The pattern is consistent across multiple independent studies. The research question is not whether it happens. It is how widespread it is, how fast it progresses, and whether citizens who recognize the pattern can reverse the trajectory before it reaches burnout.