AI Emotional Duality Index (AEDI) Dataset: YouTube Consumer Discourse on AI Fear and Curiosity — Southern European and Mediterranean Context (n=2,942, April 2026)

This dataset contains 2,942 YouTube comments collected viathe YouTube Data API v3 (April 2026) across five AI-specificdiscourse domains targeting naturalistic consumer emotionalresponses to artificial intelligence in Southern Europeanand Mediterranean contexts: (1) will ChatGPT take my job fear automation 2025(2) ChatGPT AI amazing impressive reaction review 2025(3) AI will replace humans Europe jobs scared worried 2025(4) artificial intelligence future exciting opportunity Europe(5) AI deep fake privacy surveillance worry concern 2025 The corpus is the primary empirical dataset for the AIEmotional Duality Index (AEDI) — a novel computationalmetric providing the first SML-based validation of theDual Emotional Engagement Theory (DEET) proposed byAritzis (2025) in doctoral research on AI and Greekconsumer behavior. AEDI = AI Fear / (AI Fear + AI Curiosity)1.0 = pure AI Fear dominant0.5 = balanced DEET zone (fear-curiosity equilibrium)0.0 = pure AI Curiosity dominant Three signal types — grounded in DEET's four-phase cycle:Activation → Curiosity → Resolution → Recurrence AI FEAR (AF): protective/avoidance emotional discourse Context-windowed markers (AI + fear within 20 chars) → "AI is dangerous", "scared of AI", "AI will replace us" → "AI is scary", "worried about AI", "AI surveillance" Grounded in: Rachman (1998) fear as behavioral regulator; Rogers (1975) Protection Motivation Theory; Hofstede (2001) cultural uncertainty avoidance AI CURIOSITY (AC): approach/exploration emotional discourse Context-windowed markers (AI + curiosity within 20 chars) → "AI is amazing", "love ChatGPT", "AI is incredible" → "AI game changer", "AI mind blown", "AI so cool" Grounded in: Loewenstein (1994) information gap theory; Litman (2005) epistemic curiosity; Deci & Ryan (2000) Self-Determination Theory EMOTIONAL AMBIVALENCE (EA): simultaneous fear + curiosity → "AI both scary and exciting", "mixed feelings about AI" Grounded in: Cacioppo & Berntson (1994) dual-valence affect; DEET's defining psychological state KEY FINDINGS:- Total corpus: n=2,942 comments, ~50 videos- Categorised (AEDI computed): 31 (1.1%)- Mean AEDI: 0.8226 — AI Fear strongly dominant- AI Fear cluster: 25 (80.6% of categorised)- AI Curiosity cluster: 5 (16.1%)- Emotional Ambivalence: 0 detected- AF density: 0.0545/100 tokens- AC density: 0.0119/100 tokens (4.6:1 ratio) DOMAIN-LEVEL AEDI:- AI deep fake / surveillance: AEDI=1.0000 (pure fear) AF=7.03/100 | AC=0.00 — zero curiosity markers- will ChatGPT take my job: AEDI=1.0000 (pure fear) AF=6.75/100 | AC=0.00- AI replace humans / jobs: AEDI=0.8571 AF=2.20/100 | AC=0.84/100- ChatGPT amazing / review: AEDI=0.5833 (contested) AF=5.78/100 | AC=3.83/100- AI future exciting / opportunity: AEDI=0.5000 (DEET zone) AF=3.35/100 | AC=1.50/100 CRITICAL FINDING — DEET VALIDATION:Mean AEDI=0.8226 confirms DEET's prediction that AI Fearis the dominant emotional orientation in Southern Europeanconsumer discourse, consistent with Hofstede's (2001)high Uncertainty Avoidance Index scores for the region(Greece=112, Portugal=99, Spain=86, Italy=75 — all aboveEU average of 65). The domain gradient from AEDI=1.0000(surveillance/job loss) to AEDI=0.5000 (opportunity frame)replicates the Framing Inversion pattern documented inASAI (citation omitted blind review) — confirming thatdiscourse framing, not AI exposure itself, determinesemotional orientation. Most-liked AI Fear comment (3,741):"Even the AI is saying AI is dangerous" — demonstratesthat meta-awareness of AI risk amplifies rather thanreduces fear, a finding consistent with DEET's Recurrencephase theory. Files:- aedi_v3_videos.csv: ~50 unique videos metadata- aedi_v3_comments.csv: 2,942 raw comments- aedi_v3_results.csv: AEDI scores + signal annotation Method: YouTube Data API v3. Python 3.12, April 2026.