Trust and Manipulation in Generative AI: A Digital Humanist Perspective

This paper explores the dynamics of trust and manipulation in generative AI systems, proposing digital humanism as a critical framework to re-evaluate our relationship with such technologies. We conceptualise trust as an evaluative act – a normative judgement about the trustworthiness of a system in a given context – and argue that trust in generative AI is structurally misguided. This is not because such systems lack moral agency, but because the trust placed in them has been uncritically extended from deterministic technologies, whereas generative models are probabilistic and non-linear. These systems should be approached not as ‘truth-tellers’, but as ‘storytellers.’ We further argue that deceptive features – such as their anthropomorphic linguistic style and confident rhetorical tone – exacerbate this misalignment, making users more vulnerable. Digital humanism offers a fruitful perspective for understanding these dynamics, encouraging us to engage with AI not as neutral tools, but as cultural artefacts that shape our values, behaviour, and epistemic practices.