Doing AI Differently: Rethinking the Foundations of AI via the Humanities

Artificial Intelligence is rapidly becoming global infrastructure – shaping decisions in healthcare, education, industry, and everyday life. Yet current AI systems face a fundamental limitation: they are shaped by narrow operational metrics that fail to reflect the diversity, ambiguity, and richness of human experience.This white paper presents a research vision that positions interpretive depth as essential to building AI systems capable of engaging meaningfully with cultural complexity – while recognising that no technical solution alone can resolve the challenges these systems face in diverse human contexts.We identify a foundational gap: AI systems increasingly produce and act upon cultural outputs – language, images, narratives – yet lack frameworks for interpreting the cultural content they generate and encounter. At the same time, AI is entering domains where success is harder to define: areas without clear ground truth that demand contextual reasoning and interpretive judgement. In such cases, traditional benchmarking breaks down.These interpretive challenges fall precisely within the expertise of the humanities, arts, and qualitative social sciences disciplines that specialise in understanding cultural meaning, contextual nuance, and interpretive complexity. This gap creates measurable deployment failures and ethical risks across diverse contexts, limiting AI’s effectiveness and global applicability. We’ve seen this before: early social media platforms were released with minimal contextual safeguards and benchmarked on simplistic engagement metrics – leading to unanticipated societal harms. AI, now entering even more sensitive domains, must not follow the same path. But the opportunity is greater than the problem: integrating interpretive capabilities could unlock significant advances in AI’s ability to solve complex, real-world challenges while ensuring these technologies amplify rather than erode human potential. This is a critical moment to shape AI’s foundations – early design choices will steer its trajectory for years to come.