AI – Trustworthy By Design: How to build trust in AI systems, the institutions that create them and the communities that use them

Published:
Download

There is much excitement about the opportunities of AI to improve productivity, streamline business practices, and simplify tasks. It is also very well recognised that trust in AI will be key to fully realising these benefits.

Trust, however, can be a nebulous concept. People extol its virtues and study it in surveys (according to the 2024 Edelman Trust Barometer survey the AI industry is the only sector that did not experience a year-on-year boost in trust) but there is a lack of clarity around what it means to have or lose trust and about how it is best achieved.

In this provocation paper we aim to demystify the concepts of trust in AI. We delineate trust from trustworthiness and emphasise the importance of putting trustworthiness first to fully realising the benefits of AI. We outline component elements of trustworthiness that work together to build an ecosystem of trust around and throughout the AI lifecycle – (1) AI tool reliability, (2) institutional processes, (3) meaningful stakeholder engagement – and we offer recommendations for how these components of trustworthiness can be pursued and demonstrated.