Ethical AI, Standardisation and Language Resource Management

The advent of generative AI driven by large language models has brought about a wave of impacts across all sectors in society. The swath of changes has given rise to new requirements on LLMs, which have in particular called for the alignment of AI with human values and ethical principles. In this talk, I shall address some of the emergent requirements as well as challenges and argue that, amongst the tripartite building blocks of AI, computing power, algorithm and data, language resources as the primary data hold the key to sustainable development and that standardised approaches to language resource management and annotation will offer a remedy for, among others, the data peak problem being encountered by AI systems. I shall proceed to describe the current framework of ISO Technical Committee 37 Language and Terminology and some of its latest initiatives including, in particular, a new ISO project within its working group on semantic annotation and representation to draft an international standard for the annotation of evaluative language, addressing the urgent issue of ethical AI aligned with social values.