This talk examines the gap between how AI is currently built and deployed in everyday life and what truly responsible practice in culture and society should look like. Using concrete examples from social media, workplaces, education, and public services, it maps where AI already shapes our choices and opportunities, and highlights problems like opaque decision-making, limited public input, and unclear accountability. It then introduces explainable AI in practical terms, shows how bias creeps into models and amplifies existing inequalities, and offers concrete strategies and checklists that policymakers, designers, and citizens can use to question and evaluate AI systems. Finally, it explores tensions between fairness and privacy in a ‘datafied' society, pointing out the limits of anonymization and proposing realistic compromises. Overall, the goal is to equip people and institutions with the language, questions, and tools needed to push AI toward being explainable, equitable, privacy-aware, and genuinely supportive of democratic and socially just futures.