5. Economic and Social Impact of AI

Social Level Conflict

Intellectual Property of AI

Hallucination

Harvesting Personal/Sensitve Data

Cybercrimes

Cybercrime

GenAI

• Generate new and highly coinvincing phishing emails

• Create realistic sounding speech

Cybercrime II

AI enables criminals to automate attacks, generate highly convincing social engineering, create adaptive malware, evade detection systems, corrupt training data, perform identity fraud, optimize ransomware, spread misinformation, analyse networks for weaknesses, and target critical infrastructure more efficiently.

AI Deepfakes Porn

• Showing another person’s face on another’s body

• Hard to detect and punish creators

Harvesting Data

China using facial recognition AI to monitor control and coerce its own citizen

3 Key Values to be protected

1. Safety and security

2. Ensuring diversity and inclusiveness

3. Promotion of human rights and fundamental freedoms

EU Regulations on AI

3 Categories

Unacceptable Risk - Exploitive/Social Scoring

High Risk - Biometrics/Employee Management

Limited Risk - Chatbots/Spam filters

AI Trade offs

High Predictive power vs Explainability & Transparency