AI UK 2022

Is the UK making it too hard to bring in talent or form a start-up? What seats are needed at the table to actually deliver on the people powered National AI Strategy? This panel will consider how we can put this part of the strategy into action and generate a prepared UK workforce.

AI UK co-chairs, Michael Wooldridge and Rachel Franklin, and Turing's Chief Scientific Officer, Mark Girolami will open the in the lab stage by sharing top AI discoveries that have changed the world and what's potentially to come.

Explore the future of policy-making after COVID-19: what did the pandemic teach us? Our experts discuss what technologies helped us and the potential that those technologies have moving forward. 

Reflecting on COP26, is there a vision of hope, and a surge of mobilisation for the role of AI in the path to net zero and climate change adaptation? What can we expect from AI technologies and climate science in the next 10 years and how must agencies and Institutes work together to achieve our ambitions? More speakers to be confirmed.

Hear from AI experts and Turing AI World-Leading Researcher Fellows as we explore the latest research 'from the labs'.

Dustin Carlino, Research Associate at The Alan Turing Institute will be discussing: A digital twin to design, analyse, and visualise low-traffic neighbourhoods:

Imagine a London with vehicle traffic restricted to a network of major roads, with no through-traffic anywhere else. To combat rising levels of drivers cutting through residential streets, local authorities across the UK have been creating low-traffic neighbourhoods (LTNs). Modal filters in the street physically prevent drivers from crossing, but allow pedestrians and cyclists. When filters are strategically placed to prevent all through-traffic, people in the area are likely to enjoy better air quality, less noise pollution, and higher levels of physical activity. To help design and predict the effects of LTNs, the A/B Street project has created a new open source tool, ltn.abstreet.org. In a web browser, anyone can analyze vehicle shortcuts through an area, create new modal filters, and predict any "spillover" effects onto nearby streets. Users can also check how changes affect their own personal car journeys, by comparing routes before and after a proposed LTN. This talk will demonstrate the software, then dive into the algorithmic challenges building it, such as partitioning space between major roads, calculating cell connectivity, predicting cut-throughs, and heuristics for automated filter placement.

Mirella Lapata, Professor of natural language processing in the School of Informatics at The University of Edinburgh will be discussing: Movie Summarization as a Testbed for Machine Reasoning

The success of neural networks in a variety of applications and the creation of large-scale datasets have played a critical role in advancing machine understanding of natural language on its own or together with other modalities. In this talk, we will argue that long narratives as exemplified by movies can be used to approximate real-world understanding of language and the complex reasoning associated with it. So what are the reasoning steps involved when we are watching a movie? As the movie progresses, we become familiar with the basic plot, main characters, important action, the story and the film's main themes. In the end, we may be able to produce a summary of what happened and who did what to whom. We will approximate some of these steps with a model that segments movies into thematic units, called turning points, and illustrate how they can facilitate the analysis of complex narratives, such as screenplays. We will further formalise the generation of a shorter version of a movie as the problem of identifying scenes with turning points and present a graph neural network model for this task based on linguistic and audiovisual information. Finally, we will discuss why the representation of screenplays as (sparse) graphs allows us to model various reasoning steps required for movie understanding, is interpretable, and reveals the morphology of different movie genres.

Phillip Torr, Professor of Engineering Science at The University of Oxford will be discussing: Towards Robustness for Deep Learning

Computer vision research has entered a new golden age, with applications of computer vision beginning to enter all of our lives via mobile phones, autonomous cars, identifying malignant tumours etc. The startling progress over just the past few years has been enabled by a happy confluence of the availability of “big data”, cheap memory and cheap processing in the form of GPUs leading to the re-emergence of neural networks as the dominant paradigm-the “deep learning revolution”. Deep learning now sets the state-of-the-art in nearly every area of classical computer vision, including tracking, action recognition, segmentation, object recognition, depth recovery, image synthesis, and looks to be the cornerstone of high performing methods for some time to come. However, despite this performance deep nets have also been shown to be vulnerable to adversarial examples – images which are classified incorrectly (often with high confidence), which a human would find indistinguishable from one of the correctly labelled training data. Indeed current deep learning based computer vision algorithms turn out to be surprisingly fragile to perturbation. The existence of adversarial examples challenges the trust in computer vision systems and it has been highlighted by EU’s High-Level Expert Group on AI that deployed AI should be robust. Recent computer vision related deaths in autonomous cars highlight this need. Within this talk, I will give an outline of the problem, and some directions for its solution which should be accessible to a general audience.

Explore AI for economic policy-making: looking at the experiences, challenges and opportunities in the UK. Listen to a short presentation on the labour-flow-networks project with BEIS, followed by a panel discussion on broader applications of AI for economic policy-making in the UK government. More speakers to be added.

The effects of climate and environmental change are wide-ranging; storms, flooding and other extreme weather events, increasing temperatures and polluted air- the list of environmental hazards that affect all aspects of society could go on. But how can we use data science and AI to mitigate the effects of climate change and to ensure that we are best placed to adapt to changes in our environment? Join us in this thought-provoking session to find out.

Join this years’ guest speaker, Steph Wright in conversation with Anjana Ahuja, as together they explore what they’ve been reading and thinking about recently on diversity in AI, with a particular focus on intersectionality. We’ll also hear more on Scotland’s AI strategy.

A panel of entrepeneurs and start-up leaders talking to their former selves about their top tips and lessons learned. Speakers to be confirmed.

Session details coming soon.

Globally, states are utilising AI to develop and maintain strategic advantage over adversaries. However, efforts to develop and employ new AI capabilities outpace our ability to establish norms for their use. On this panel, we consider what the use of AI means for issues of global security and international peace across increasingly fraught global politics, including existing capabilities and how these challenges can be best addressed.

Session details coming soon.

In this session we introduce the Online Harms Observatory, a new analytics platform from The Alan Turing Institute’s public policy programme. It combines large-scale data analysis and cutting-edge AI developed at the Turing to provide real-time insight into the scope, prevalence and dynamics of harmful content online. It aims to help policymakers, regulators, security services and civil society stakeholders better understand the landscape of online harms. The OHO is supported by DCMS. We will outline why the Observatory is needed and what it aims to achieve, and will showcase the first tracker – a live monitor of abuse directed against Premier League football players on Twitter.

Join us as we hear from experts in the field of AI and biodiversity, including Professor Nick Wright, who will explore 'What can dolphins tell us about ocean health?'.

Inspired by the ODI Summit – join in with another episode of the 'data game' and see how these AI figures fair testing their data know-how!

Hear from a selection of leading researchers and experts covering topics such as synthetic data, differential privacy, homomorphic encryption, federated analytics. 

Online platforms have opened up unprecedented avenues for people to connect, interact and access information. However, they have also created a clear risk of harm, from child abuse to hate speech; misinformation to suicide and self-harm. After years of 'self-regulation' a spate of new laws have been introduced across the globe, not least the UK’s Online Safety Bill. It aims to hold platforms to account for how content and activities that could inflict harms on people online disseminate on their service, empowering the new online safety regulator, Ofcom, to act when failures are identified.

In this panel we bring together Damian Collins (Chair of the Draft Online Safety Bill PLS in UK Parliament), Anna-Sophie Harling (Principal in Online Transparency at Ofcom), a speaker from DCMS and from 5 Rights. We will discuss what the new online safety regulation will mean in practice, what else needs to be done to ensure online safety, and whether the regulation is fit for handling new tech, such as augmented and immersive services like the Metaverse. The panel is hosted by Dr Bertie Vidgen, Head of Online Safety at the Turing.

Join us for our climate-focused AI in action demonstrations, showcasing cutting-edge AI that's being used to solve real-world problems in organisations and labs throughout the UK.

The Turing has recently become a supporter of the Better Images of AI project – a movement to improve the representation of AI. Open to all, this is a special advisory session for all those with an interest in communicating about AI. Tania Duarte of We and AI, Dr. Kanta Dihal of the Leverhulme Centre for the Future of Intelligence, and Tristan Ferne of BBC R&D will explore the many positive effects of more realistic visual representations of AI.

Hear from Turing AI Acceleration Fellows discussing their latest projects:

Professor Aldo Faisal will explore How Human AI will need neuroscience more than neural networks:

AI is revolutionising the world, but how can we harness its power for changing how we spread its benefits? In our lab, we focus on the interplay between human and artificial brains, what they have in common, and most importantly how we can use principles from neuroscience to improve technology and use frameworks from AI to advance our understanding of the neuroscience of human-AI interaction. We will showcase two examples highlighting how robotics and AI can be used to co-design novel solutions to existing challenges.

Professor Damien Coyle will discuss AI-enabled neurotechnology:

Research in the field of brain-computer interfaces (BCIs) and neurotechnology has proven that electrical signals in the brain, modulated intentionally by mental imagery, can relay information directly to a computer, where it is translated by intelligent algorithms into control signals that enable communication and control without movement or can improve self-regulation of brain activity. This talk will present results that show people with restricted abilities resulting from disease, injury or trauma may benefit from neurotechnology, including those who have prolonged disorders of consciousness or locked-in syndrome following traumatic brain injury, spinal injury, stroke and post-traumatic stress disorder. Neural activity can be modulated by many kinds of mental imagery e.g., classical motor imagery BCIs distinguish between imagined hand/arm movements. In this talk, recent results in decoding imagined three-dimensional limb movements and silent/imagined speech from EEG will all be presented. The presentation will attempt to address the question is it feasible to expect high and robust performance with these types of imagery in EEG-based BCIs and will highlight results that indicate user proficiency in BCI control is a matter of training time, machine learning/AI ability, application of the technology and maintenance of stable affective states.

Professor Yulan He will present her latest research on Machine Reasoning for Natural Language Understanding:

Enabling machines to understand human languages requires equipping AI systems with reasoning capabilities. Reasoning is not necessarily achieved by making logical inferences. In our context, it refers to being able to manipulate previously acquired knowledge in order to enhance the understanding of new information encountered. In my talk, I will outline our recent progress in machine reasoning, including incorporating commonsense knowledge for emotion detection in dialogues, event semantic reasoning for answering questions relating to real-world events reported in news, and evidence reasoning for rumour veracity assessment.
 

Launch of the Modelling for Policy Theme followed by a panel on the role of computational social science (CSS) for policy-making. The expert panelists are world-leading academics who have made seminal contributions to CSS. More speakers to be added.

Hear from the Chief Scientific Officer of the UK government as we ask - how does the National AI Strategy fit in to the overall science and technology strategy for the UK? What have we learned from the pandemic, and what can AI offer us to be more resilient and efficient in the future? 

Join Anjana Ahuja in conversation with Sana Khareghani and Tabitha Goldstaub, exploring the progress of the National AI strategy. Six months after its launch, find out what's been successful, what's currently being prioritised, and learn what areas you need to be leading on, as the UK enters a crucial phase in the strategy.

As the levelling up agenda develops, are more opportunities for growth coming to local and regional governments? A panel of of regional experts and explore how AI might fit into their ambitions and what role AI will play in improving their cities infrastructures and services. But what are the challenges to address from skills gap to data inequalities?

The PhD is a fundamental milestone on the path to become a scientist – but it can be a challenging experience. How can you decide whether a PhD (and which PhD) is the right next step for you? How do you choose a fulfilling topic? And how can PhD students develop resilience to take failure in stride and keep growing?

Dr Victoria Volodina, Clinical Operations Research Unit (CORU) UCL will be discussing: Uncertainty Quantification (UQ) methods for calibration of climate models.

Tuning in the climate modelling community is crucial in ensuring that the model is a trustworthy representation of the physical process of interest. In this talk, I will briefly discuss the importance of emulation and history matching and illustrate the performance of these UQ methods in application to climate model tuning.

Dr Louise Kimpton, Postdoctoral Research Fellow, University of Exeter will be discussing: Uncertainty Quantification for Multi-level Models

Complex computer codes can sometimes be run at different levels of complexity or resolution. Using well known techniques, we can fit Gaussian process emulators (GPs) in order to better understand the models and make predictions. The locations of the input parameters of these models greatly effects the accuracy of these GPs, and so the input design becomes an important consideration. Given an initial design and associated GP, a sequential design selects the optimum next parameter setting to reduce the error on the GP. In this talk, we will briefly discuss a method for sequential design for multi-level models. Not only must we select the optimum input parameter location, but also the model level in which to run it at.

As highlighted in the National AI Strategy, technical standards are set to play an important role for advancing AI technologies and ensuring their effective governance. The AI Standards Hub is a new initiative dedicated to community building, knowledge sharing, and international engagement around AI standardisation. Join this fascinating session to learn more about the AI Standard Hub’s mission, and listen to experts from the organisations partnering to pilot the Hub as they talk about their work behind the scenes and discuss the significance of technical standards for AI in the UK and internationally. 

How do you measure progress in AI? What data is available and what sources can we rely on? This panel of scientists will debate how we quantify scientific progress and if this benefits society. What is the GDP equivalent?

Innovation in AI and data science is revolutionising our ability to diagnose and treat diseases.

The earlier a disease is detected, the better the chance of successful treatment for patients. For example, more than nine in 10 bowel cancer patients survive the disease for five years or more, if diagnosed at the earliest stage.

Join us as we explore the latest innovations in using AI to diagnose diseases, and we reflect on what the future might hold for AI-driven health technologies. Anyone with an interest in the application of AI to improve patient outcomes is encouraged to join us.

From talks on battery development to materialism manufacturing – let's explore how AI is transforming how we manage energy. Listen to three groundbreaking research talks with time for Q&A.

Artificial intelligence is a general-purpose technology that touches all areas of our economies and societies. It's changing businesses by streamlining internal operations, contributing to product development, improving marketing and sales strategies, and reshaping risk management. It's also changing our societies by curating the news stories that we read online, scanning our faces to verify our identities, matching us with jobs and partners, translating foreign-language content into our native languages – to name only a few of its uses. AI is a game-changer and will have a profound effect on our lives, communities, and economies.

Against this backdrop of AI innovation, the popular imagination has been captured by stories of robots and super-intelligent agents taking over the world. These stories in which humans are helpless victims of technological innovation ignore the fact that we are in the driving seat of these advancements. Regulation is a key component of the arsenal of tools that we have to shape technological innovation so that it is to the benefit of humanity. Regulation sets the guardrails for this journey that we are on, ensuring that we accrue the enormous benefits that AI brings while minimising current and future harms.

This session will highlight the role of regulation in guiding AI innovation. It will explore the opportunities and barriers to developing well-crafted laws and sensible regulation and shed light on the need for international cooperation in this space. We will start with an introductory address by Professor Julia Black, President of the British Academy as well as Strategic Director of Innovation and Professor of Law at the London School of Economics and Political Science. We will then open the floor to a panel discussion led by Dr Florian Ostmann, who is Head of Governance and Regulatory Innovation at The Alan Turing Institute.

This talk will introduce the Surrey Institute for People-Centred AI and present research on the AI4ME partnership with the BBC, introducing the use of AI for personalised media content, production and delivery. Personalised media enables content to be tailored to the individual interest, accessibility requirements, device and location. The research will be discussed in the context of AI for education and entertainment.

Our health and well-being are intrinsically linked to the environment in which we live and, as our climate changes, living healthily may be increasingly more challenging. In this session, we take a deep dive into the changes in temperature and air quality that are affecting our health and that of generations to come.

Join renowned IT Entrepreneur and Philanthropist Dame Stephanie Shirley for an opening address, followed by a fascinating conversation hosted by Judy Wajcman, lead on the Women in Data Science and AI project at The Alan Turing Institute. In a wide-ranging discussion, expect to learn about Dame Stephanie's inspiring journey from an unaccompanied child refugee to the founder of a software business ultimately valued at almost $3 billion, her advice on how women and underrepresented groups can find a home in STEM, and her philanthropic work and pioneering projects in autism.

Join Professor Sarah Dillon in conversation with Jeanette Winterson, exploring her new work, 12 Bytes: How Artificial Intelligence Will Change the Way We Live and Love, and exploring the future of AI with challenging questions about humanity, art, religion and the way we live and love.

The New Real Observatory invites leading digital artists to experiment with environmental datasets, climate models and processing pipelines to fuel a new generation of technology-conscious aesthetics situated in our connection with land, water, air and energy. 

Behind our AI-powered technology lies an invisible workforce. Thousands of ‘ghost workers’ perform tasks such as cleaning and labelling data, recruited via crowdsourcing platforms for as little as $2 per hour. With increasing concerns about exploitation, is this the digital equivalent of fast fashion sweatshops?

The past shapes the present and influences the future, but the historical record isn’t straightforward, and neither are its digital representations. Join the AHRC project Living with Machines and friends on their journey to reimagine the past through AI and data science and the challenges and opportunities within.

What are the strengths and weaknesses of foundational models such as GPT3? Are they the future and what are their limitations and societal implications?

Experience a taste of The Alan Turing Institute's Data Study Groups (DSGs). All afternoon, in a sprint DSG-style event, volunteer researchers will be collaborating in groups, applying data science tools and methodologies to a climate change challenge.

Join this panel discussion to hear representatives from each group discuss how different disciplines and experience levels came together to work towards a common purpose. What was surprising? What did they discover and how did they work together? Uncover the DSG way of doing research and let us inspire you to join our unique training, research-led events.

Over the past couple of years, there has been increasing interest in global and intercultural approaches to AI ethics and governance. A good example of this is the publication of UNESCO’s Recommendations on the Ethics of AI, adopted in the fall of 2021 by its 193 member states. Despite growing momentum, proponents of this new policy direction face a set of difficult challenges. For instance, the development of global AI ethics frameworks has thus far been largely dominated by Western and Global Northern voices, making the active inclusion of non-Western as well as decolonial and postcolonial perspectives ever more essential. Likewise, international AI policy, standards, and regulatory environments have increasingly been shaped by power asymmetries that have arisen from the control of digital infrastructures by large multi-national tech corporations and high-income countries. In this panel, we will explore some of these thorny issues. Our conversation will cover both areas in need of critical attention and prospects for technology policy innovation with a view to the importance of the intercultural dimension. 

Take top academics, contentious research, add a comedian and it’s the force of nature that is the Cabaret of Dangerous Ideas (CODI).

Now in its tenth year, CODI drags academics out of their ivory towers to debate today’s burning issues and cutting edge research. Live from The Stand in Edinburgh.