Natural Language Processing (NLP) is an area of AI that has rapidly gone from technical experimentation in labs to large-scale deployments in-the-world. Each day there are new stories about the transformative potential of applications that are underpinned by NLP. For example, claims that ChatGPT will change the way we produce new texts; or that the next generation of robo-advisers and chatbots will revolutionise service industries; or that Alexa and Siri will enable hands-free access to information through voice; or that text-to-image software like Midjourney will transform, and deeply threaten, the creative industries.
NLP is not just at the vanguard of applied AI but also at the forefront of questions around trust and responsibility. When NLP applications are placed in-the-world, they come into direct contact with peoples everyday lives, get used in unexpected ways, and incur interactions that were not modelled in the lab. As NLP gets deployed outside the lab, many questions are raised about not just what can be done, but what SHOULD be done, with these technologies - with continued concerns around the ethics of language data, biases in datasets and models and associated harms, and whether new applications may take away human agency and autonomy. Because of how NLP sets out to utilise, and mimic, human language and communication, it poses especially complex socio-technical, design and legal challenges, and raises public debate about the implications of AI more broadly on society.
Our CDT will train 50+ future leaders that represent a new generation of interdisciplinary experts that can ensure NLP systems are designed, implemented, deployed and adopted in responsible ways that are people-centred and trusted by society. We have established a new training programme that supports students to engage in cutting-edge research and practice related to the technical, social, design, and legal and regulatory challenges these technologies present. Core to our training is ensuring students come out with the required skills to co-create novel applications with users and are ready to collaborate in the interdisciplinary teams that will be needed to address complex socio-technical challenges. Fundamentally, we seek to transform how these technologies are developed, to ensure the concerns and needs of users, and governance and regulatory requirements, are integrated into the upstream technical development of NLP systems, rather than left as downstream and separate concerns.
The CDT provides a unique training environment by bringing together 70+ academics across the University of Edinburgh who are extensively engaged in NLP and human-centred data and AI research: (i) experts in the technical foundations of NLP and language models within the Institute for Language, Cognition and Communication; (ii) experts in human-computer / human-AI interaction, visualisation and interaction design, and technology ethics and regulation design, within the Institute for Design Informatics; and (iii) experts in speech technologies in the Centre for Speech Technologies. Our students will benefit from being co-located in dedicated facilities within the Edinburgh Futures Institute; the University's new £170m data-driven innovation hub focused on interdisciplinary, challenge-based research and societal impact.
Embedded throughout is close collaboration with 60+ industrial, public and non-profit partners. Partners provide contributions to the training of students by setting contexts and problems to work on, by hosting internships and groups of students working on Applied Innovation Projects, and offer non-academic career pathways for graduates. Our partners include leading national and international organisations with a direct interest in building research capacity in trustworthy and responsible NLP across diverse sectors, e.g.: Cohere, Microsoft Research, BBC R&D, DCMS, NatWest, abrdn, Public Health Scotland, Codebase, ScotGov, Jaguar Land Rover.
|