Dr Joost Noppen (BT)

Rolling out Generative AI for Software Development in BT: experiences and lessons learned

Generative artificial intelligence has been in all the headlines for quite some time now and during this time we have seen gen AI tools emerge that can assist with almost task. In particular within software development a plethora of tools is on offer, and with claimed productivity benefits of 40% or more one can be forgiven to think that rollout within an organisation is a bit of a no-brainer. Unsurprisingly though, the devil is in the details. Making sure you have the right tools, training and rollout to teams that can take advantage of the capabilities in their projects is far from trivial. And understanding its impact on productivity and value created can be very challenging as this is rarely as easy as counting lines of code generated. In this talk we will examine the rollout of Code Whisperer in BT, what went right, what went wrong and what we have learned for the future.

I am chief researcher in software engineering, leading a team of researchers focussed on the future of software engineering in BT. Our primary focus is on understanding the impact and leveraging of novel approaches and technologies such as artificial intelligence on the practice of software development, and propose new ideas and tools together with academic and industry research partners. I hold an MSc and PhD in Computer Science from the Universiteit Twente (Netherlands), both with a specialisation in Software Engineering.

Prof Ashiq Anjum (University of Leicester)

Self-Learning Digital Twins for Cyber Physical Systems

Digital Twins bring together cyber and physical systems in a single platform that continuously evolves as the underlying physical systems change over time. This talk will present system models and approaches in building digital twins that are physics inspired and intelligently assimilate physics models and AI algorithms into a single unified platform. Through example projects and applications, the talk will highlight advancements in systems, models and algorithms that have contributed to self-adapting digital twins and how emerging applications will necessitate networks of digital twins that adapt as individual physical components within a networked distributed system evolve and learn from the system behaviour. The talk will provide insights about the current advancements in Physics informed AI that are leading us to a future where networks of digital twins will become the mainstream cyber-physical systems. The talk will conclude with the challenges and opportunities that are driving the evolution towards self-adapting networks of digital twins and how distributed physics informed AI could help us in realizing this vision.

I am a professor of Distributed Systems at the University of Leicester. I am also the director of enterprise and impact and previously have been the director of the data science research centre. My areas of research include self-learning digital twins for cyber physical systems, distributed machine learning models for self-adapting networks of digital twins and physics informed machine learning models for digital twins to emulate the real time behaviour of networked cyber-physical systems. I am leading the Digital Twins initiative with British Telecom that is producing digital twins that can emulate the complex functionality of large and complex telecom networks. I am also involved in the SEEDS programme with CGI that is producing Digital Twins for sustainable Data Centres. I am also leading the digital twin development in the EPSRC funded project “Self-Learning Digital Twins for Sustainable Land Management” as well as in the EPSRC funded project “Self-Learning AI-Based Digital Twins for Accelerating Clinical Care in Respiratory Emergency Admissions (SLAIDER)”. We are also developing digital twins for cosmic explorations in the EPSRC funded project “Blueprinting AI for Science at Exascale - Phase II (BASE-II). In addition, I am also involved in other digital twin initiatives such the Cardiac Digital Twin with cardiovascular sciences as well as Thermal Digital Twins with aerospace industry. My research work has been funded through a number of research grants including the EPSRC projects in AI driven digital twins for net zero and clinical care and the EU funded projects on distributed clinical intelligence, modelling and iterative genome analytics. I have been investigating distributed analytics platforms for the LHC data in collaboration with CERN Geneva Switzerland for the last twenty years. I am also actively collaborating with leading VR/AR companies to commoditise AI driven digital twins to translate real time physics informed and distributed machine learning algorithms for augmented and virtual reality based digital twins with applications in health, aerospace and telecom.

Dr Salvador Pacheco-Gutierrez (United Kingdom Atomic Energy Authority)

Data centralisation platforms to support planning and decision-making in nuclear decommissioning

The UK government prioritises cleaning up legacy nuclear sites safely, securely, and cost-effectively. This mission now includes decommissioning and repurposing the Joint European Torus (JET) starting in 2024. A major challenge in this effort is the sparsity of information due to decades of evolving data management technologies, confidentiality concerns, and other practices. This lack of comprehensive and current data complicates the decommissioning process by hindering accurate assessments of facilities, identifying hazards, evaluating structural integrity, and estimating costs. Addressing this issue requires extensive data collection and thorough digital documentation to ensure decommissioning plans are based on accurate and complete information. A well-structured data set will enable the use of advanced AI technologies to support forecasting, prevention, and optimisation of decommissioning scenarios.

Dr Salvador Pacheco-Gutierrez is the technical lead of the Robotics and AI data theme within the Robotics and AI Collaboration (RAICo) program, a collaboration between the NDA, UKAEA, Sellafield LTD and The University of Manchester. His background is in robotics, computer vision, software, and AI. He has participated in multiple research and industrial projects on robotics and its applications. His work within RAICo aims at leveraging legacy and newly collected data from nuclear sites to facilitate decision-making by developing a centralised, interactive, smart, and efficient source of truth to support decommissioning planning.

Dr Ciprian Zavoianu (National Subsea Centre, Robert Gordon University)

Supporting the energy transition with AI tools for optimal trade-off discovery

The coupling of state-of-the-art non-linear optimisation techniques with efficient scenario-dependent operational simulations improves the ability to quickly explore the design space of various energy-centric projects, providing domain experts and decision-makers with a set of optimal trade-off solutions suitable for secondary in-depth analysis. Whilst usually requiring high-performance or high-throughput computing resources, this AI-driven search for optimality generally improves attainable KPIs and fast-tracks new developments related to energy transition targets by driving an overall reduction of design times. Over the years, we have successfully applied this generic simulation <=> optimisation approach to tackle complex real-life problems at component level (e.g., designing robust electrical drives), asset level (e.g., optimising wind farm cable layouts), and system level (e.g., maximising the impact of introducing low emission vehicles in existing public transport systems).

Dr Ciprian Zavoianu leads the Net Zero Operations research programme in the National Subsea Centre (Aberdeen) and his research interests for the past 10+ years have included theoretical and practical concepts of Artificial Intelligence (AI). In particular, his research is primarily concerned with evolutionary computation algorithms and their application on complex real-world problems that have solutions predicated on combining simulation, optimization and data-driven modelling. Over the years, Ciprian’s research has been applied internationally for (i) improving hi-tech manufacturing processes, (ii) discovering optimal trade-offs in the design of various electrical and mechanical assemblies, (iii) modelling and optimising multi-modal transport networks. He has published over 40 peer-reviewed articles in leading AI conferences and journals including: GECCO, PPSN, CEC, Knowledge-Based Systems, Applied Soft Computing, Engineering Applications of Artificial Intelligence, Information Sciences and Transactions on Evolutionary Learning and Optimization.

Sabeehah Mahomed (Alan Turing Institute)

Children and AI

Children of all ages are increasingly interacting with AI in their daily lives, including through playing with smart devices or toys, engaging with online content or through the use of services where AI systems inform decisions about children and their families. While children are typically underrepresented in decision-making around AI, AI presents specific opportunities and challenges for children. Bridging the gap between theoretical considerations of children’s interaction with technology and empirical insight into their experiences, opinions, aspirations, and questions about AI, Sabeehah will present her team’s research to advance child-centred approaches to the design, development, deployment, and governance of AI and to examine the ways that AI impacts children’s rights and wellbeing, and how children’s rights can be protected in a digital world, as well as developing and testing approaches to meaningfully involve children in decision-making relating to AI.

Sabeehah Mahomed is a Researcher for the Ethics & Responsible Innovation team in the Turing’s Public Policy Programme and Visiting Researcher at the Digital Environment Research Institute, Queen Mary University London. She currently works across diverse projects on the ethical and responsible design, development, and deployment of AI systems. In particular, this includes researching and analysing the context of children’s rights and AI through a series of engagements with children. For instance, Sabeehah has worked in collaboration with the Scottish AI Alliance and Scotland’s Children’s Parliament over a two year period engaging children across Scotland. This work was featured at the Scottish AI Summit ’23 & ’24, AI UK 2024, the QMUL/Turing AI Fringe event, and across various media platforms in the UK. Sabeehah’s work has also received international interest where she was invited by the Finnish Digital Agency to co-facilitate a workshop on children and AI in Helsinki with various stakeholders, and has worked with the Council of Europe’s Committee on Artificial Intelligence (CAI) and the Council of Europe’s Steering group for the Rights of the Child (CDENF) to design, develop, and implement a mapping study exploring the extent to which children’s rights have been considered in the context of AI across member states. In addition to her work on children and AI, Sabeehah has worked with UNESCO to co-develop the first Global AI Ethics and Governance Observatory and is a co-author of the Turing’s AI Ethics and Governance in Practice Workbooks designed for the public sector. Sabeehah holds an MSc in Digital Humanities (distinction) from the Department of Information Science/Studies at University College London (UCL) and was awarded the UK's Women of the Future (Commonwealth) Award in 2021 and is a FCDO Chevening alumni.

Dave Yearling (Senior Research Manager, Service Analytics, BTIIC)

Health at Home

We are facing a crisis in providing elderly care within the UK, with over half of requests for such care being refused. This is a product of both staffing and sheer numbers requiring care. Currently, around 19% of the population is aged 65+, with around 40% of those having at least one chronic health condition. Combined with a prolonged period of low birth rate, this is set to increase dramatically to around 24% of the population in 2040. Enabling a strong digital support system is key to addressing this. This project uses AI to provide a means for relatives, carers and clinicians to remotely monitor elderly relatives and patients in a discrete manner. It always provides the elderly person with complete control and always respects their privacy. Passively, without the need for wearables, it detects chronic changes in behaviours and mobility as well as identifying acute changes such as falls, injuries and even risky behaviours. This work is part of a wider suite of activities within our Ireland Innovation Centre collaboration with Ulster University. The centres mission is to accelerate research opportunities into well engineered realities that provide real value to the community.

In a nutshell, I am passionate about solving real business problems using statistics, operational research and data science. This means I am always seeking to get the most effective and efficient solution, regardless of hype or fashion. It’s getting the solution that matters, not the tools that we use. Currently I lead a research team looking at a variety of problems facing our business. This includes such diverse applications as privacy centric health monitoring, autonomous service orchestration, capacity planning, future of work and forecasting call volumes. This often leads the adaptation and application of a whole suite of techniques and combining them to create value. Most recently, these include anomaly/change detection, causal AI, deep learning, reinforcement learning, simulation and even traditional statistics! My background is quite eclectic. Initially a mechanical engineer, I actively took the opportunity to specialise in statistics and operational research. This has given me the opportunity to follow a wonderful career where I have been lucky enough to work on many different areas. Beginning with epidemiology I then went onto work on logistics, software development, finance, customer service, central government and now in telecommunications and networks. I hold 2 degrees, published many papers and filed numerous patents.