Fixed Term Research Fellow in Artificial Intelligence (27 months)

Job Summary

We are looking for a full-time postdoctoral research fellow for 27 months. You will be working with Dr Mark Snaith on the EPSRC-funded project Dialogue-based Structured Conversational Artificial Intelligence.

Systems powered by conversational Artificial Intelligence (AI) have seen a significant increase in uptake in recent years. The relatively recent widespread launch of platforms underpinned by Large Language Models (LLMS), such as ChatGPT, has piqued public interest in more advanced conversational systems that, ostensibly, exhibit understanding and can sustain longer form conversations. However, while such platforms have proven effective as a tool for seeking information and generating ideas, the conversations are open and unstructured. Without further scaffolding, this limits the usefulness of language models in domains and applications that require a rigid dialectical interaction structure. Providing this scaffolding in a way that is as accessible to developers as ChatGPT (and other LLMs) is an essential next step if the full potential of conversational AI in focused domains is to be realised.

Implemented dialogue games have been shown to support engaging and interactive conversational applications in diverse domains that require careful consideration of dialogue flow, such as health care, law, and dispute mediation. However, creating these dialogue game implementations is time-consuming and requires specific expertise both in dialogical analysis, and dialogue game implementation languages. If the use of structured models of dialogue in supporting conversational AI is to reach its full potential, a vital next step is to harness advances in deep learning, language models, and argument mining to lower these barriers to entry.

The aim of this project is to create for the first time the theories, tools and techniques that will support widespread use of structured models of dialogue in underpinning complex domain-specific conversational systems. By building on advances in argument mining, the project will facilitate automated implementation of computational dialogue games from natural language transcripts of real examples, and develop new methods for more natural user interactions.

In keeping with RGU's commitment to work/life balance you will have the opportunity to work from home for a portion of the working week and also benefit from a generous pension scheme, 46 days annual leave (including statutory days) an onsite nursery and sports centre, as well as a range of voluntary health and travel benefits.

We are committed to attracting and engaging a diverse range of staff and fostering a culture where everyone feels welcome, comfortable to express their ideas, and can reach their full potential. For more information, please visit Our Diverse Workforce pages.

Job Description

RESPONSIBLE TO: Dr Mark Snaith

RESPONSIBLE FOR: No Supervisory Responsibilities

PURPOSE OF POST: To develop the theories, tools and techniques that will support widespread use of structured models of dialogue in underpinning complex domain-specific conversational systems.

PRINCIPAL DUTIES:

• Investigating and developing appropriate data annotation schemes to support automated extraction of dialogue structures.
• Developing appropriate theories, tools and techniques in machine learning to automate the extraction of dialogue structures from natural language text.
• Liaising with industrial and academic partners to support the deployment of practical results in usable, real-world applications.
• Writing high quality research outputs to disseminate project results.
• Presenting project results at conferences.
• Carrying out high quality research into dialogue systems, natural language processing, and related technologies.
• Writing high quality publications for top conferences and journals.
• Presenting research results, both internally to colleagues, and externally at international conferences.
• Liaising and working with the project’s industry partner(s) to deploy results and outputs into real-world applications.

Person Specification

ESSENTIAL REQUIREMENTS

Qualifications and Professional Memberships
PhD in appropriate subject

Knowledge
Excellent software development skills.

Machine learning in the context of conversational AI systems.

Evidence of high-quality research publications.

Natural Language Processing, Large Language Models.

Experience

Experience in continual learning in Large Language Models.

Experience of writing high quality research publications.

Experience of working with industry on research projects involving Natural Language processing (or closely related technologies).

Experience in machine learning.

DESIREABLE REQUIREMENTS

Knowledge

Knowledge of formal dialogue and argumentation, and in particular argument mining.

Experience

Supervision (formal or informal) of undergraduate and taught postgraduate students.

Behaviours

Behaviour 1: Communication - Ability to receive, understand and convey information requiring careful explanation and information of a complex or conceptual nature, in a clear and accurate manner

Behaviour 2: Analysis and Research - Experience of developing hypotheses and concepts to explain data, events and phenomena, and reporting findings to a wider community

Behaviour 3: Pastoral Care and Welfare - Experience of calming and reassuring those with work/study related problems who may be experiencing distress and dealing with difficult welfare situations or confidential matters

Behaviour 4: Liaison and Networking - Experience of circulating information in an accurate and timely manner, working across team boundaries to build and strengthen working relationships and leading and developing internal networks to pursue a shared interest

Behaviour 5: Initiative and Problem Solving - Experience of using initiative and creativity to resolve problems, identifying practical and suitable solutions.

Behaviour 6: Decision Making - Experience of using own judgement to make decisions, making collaborative decisions with others to reach conclusions and providing advice or information that will influence the decisions of others