Comments: 2872 ============================================================ AI HABIT ENTRENCHMENT INDEX (AHEI) v2 n=2872 | Categorised=66 ============================================================ Mean AHEI: 0.9091 (1.0=Habit Entrenchment | 0.5=Contested | 0.0=Deliberate Choice) Density per 100 tokens: Habit Entrenchment: 0.1212 Deliberate Choice: 0.0039 Dependency Anxiety: 0.0026 Signal Distribution: signal_type Uncategorised 2803 Habit Entrenchment 60 Deliberate Choice 6 Dependency Anxiety 3 Name: count, dtype: int64 AHEI by query domain: AHEI=1.0000 | HE=8.0799 | DC=0.0000 | DA=0.0000 | n=8 >>> AI dependency problem can't work without it anymore 2025 AHEI=1.0000 | HE=4.5288 | DC=0.0000 | DA=0.0000 | n=11 >>> AI tools became part of my life routine every day ChatGPT 2025 AHEI=0.9375 | HE=6.1607 | DC=0.1252 | DA=0.0000 | n=32 >>> ChatGPT addiction can't stop using it daily habit 2025 AHEI=0.8000 | HE=3.4975 | DC=0.7692 | DA=0.0000

This dataset contains 2,872 YouTube comments collected viathe YouTube Data API v3 (April 2026) across five discoursedomains targeting naturalistic consumer expression ofhabitual, deliberate, and dependency-anxious AI use: (1) ChatGPT addiction can't stop using it daily habit 2025(2) AI tools became part of my life routine every day 2025(3) worried relying too much on AI losing ability to think(4) I use ChatGPT for everything automatically 2025(5) AI dependency problem can't work without it anymore The corpus is the primary empirical dataset for theAI Habit Entrenchment Index (AHEI) — the firstcomputational metric measuring the transition fromconscious to habitual AI use in naturalistic consumerdiscourse, addressing the habit formation gap identifiedin Cogent Business & Management (2025) as underexploredin the AI consumer behavior literature. AHEI = Habit Entrenchment / (Habit Entrenchment + Deliberate Choice)1.0 = pure Habit Entrenchment (automatised AI use)0.5 = UTAUT2 contested zone0.0 = pure Deliberate Choice (conscious AI evaluation) Three signal types — grounded in UTAUT2 and beyond: HABIT ENTRENCHMENT (HE): automatic, routinised AI use → addicted, can't live without, every day, second nature, automatically, without thinking, hooked, habit, reflex Grounded in: Venkatesh et al. (2012) UTAUT2 habit construct; Limayem et al. (2007) IS habit continuance; Wood & Runger (2016) habit as automatic behavioral script DELIBERATE CHOICE (DC): conscious, intentional AI use → I choose, when I need, just a tool, selectively, my own thinking, I can stop, balance, consciously Grounded in: Davis (1989) TAM — utility evaluation; Ajzen (1991) Theory of Planned Behaviour DEPENDENCY ANXIETY (DA): fear of cognitive over-reliance → brain rot, losing my ability, too dependent, can't think without it, deskilling, cognitive decline THEORETICALLY NOVEL: absent from all TAM/UTAUT models. First computational operationalisation of Seligman's (1972) learned helplessness applied to AI tool use. Reported separately as modifier signal. KEY FINDINGS:- Total corpus: n=2,872 comments, ~50 videos- Categorised (AHEI computed): 66 (2.3%)- Mean AHEI: 0.9091 — Habit Entrenchment strongly dominant- Habit Entrenchment cluster: 60 (90.9% of categorised)- Deliberate Choice cluster: 6 (9.1%)- Dependency Anxiety cluster: 3 (4.5%) — small but novel- HE density: 0.1212/100 tokens- DC density: 0.0039/100 tokens (31:1 ratio)- DA density: 0.0026/100 tokens DOMAIN-LEVEL AHEI:- AI dependency / can't work without: AHEI=1.0000 HE=8.08/100 | DC=0.00 — pure entrenchment- AI daily routine / part of life: AHEI=1.0000 HE=4.53/100 | DC=0.00 — pure automaticity- ChatGPT addiction / daily habit: AHEI=0.9375 HE=6.16/100 | DC=0.13/100 — entrenchment dominant- ChatGPT for everything automatically: AHEI=0.8000 HE=3.50/100 | DC=0.77/100- Worried relying too much / losing ability: AHEI=0.7000 HE=1.89/100 | DC=0.35/100 — LOWEST (most deliberate) Only domain showing Dependency Anxiety signal CRITICAL FINDING:Mean AHEI=0.9091 confirms that consumers who discuss AIhabit and dependency in naturalistic discourse expressHabit Entrenchment almost exclusively (90.9%) — withvirtually no Deliberate Choice language. HE:DC ratio=31:1.The most-liked comment (5,096): "I'm giving away too muchinformation but I can't stop cause I love talking aboutthis" — ironic self-referential entrenchment. Second most-liked (4,581): "Imagine knowing most of society has anaddiction proven to be unhealthy and doing nothing" —collective entrenchment framed as social crisis. The DAcluster (3 comments), though small, provides firstcomputational evidence of Seligman-type learned helplessnessin AI tool use discourse. Files:- ahei_v2_videos.csv: ~50 unique videos metadata- ahei_v2_comments.csv: 2,872 raw comments- ahei_v2_results.csv: AHEI scores + signal annotation Method: YouTube Data API v3. Python 3.12, April 2026.