The Turing Podcast

In an era of increasingly unpredictable extreme weather, getting the forecast right is no longer just about what you need to wear during your morning commute. It’s about national resilience, food security, and technological sovereignty.

In this episode, host Amelia Jabry is joined by Dr Scott Hosking (Mission Director for Environmental Forecasting at the Alan Turing Institute) and Professor Kirstine Dale (Chief AI Officer at the Met Office) to introduce FastNet: a groundbreaking, home-grown AI weather model developed right here in the UK.

Together, they demystify how FastNet uses 3D geometry (specifically, an icosahedral mesh) to map our round planet, how it spots patterns across 40 years of climate data. They also explore why we still desperately need physics to prepare for unprecedented ‘Grey Swan’ climate events like a looming Super El Niño. They tackle the vital questions of making our AI sustainable, the critical importance of a diverse AI workforce, and why the UK must maintain its own independent weather modelling capabilities rather than relying on global tech giants.

Read more about FastNet here: https://www.turing.ac.uk/research/research-projects/fastnet

 

Chapter markers

0:00 Intro

1:04 Introducing Scott Hosking and Kirsten Dale

1:50 The Raucous Revolution in weather forecasting of 2022

04:17 how AI weather prediction models work and how they differ from traditional weather prediction models

05:15 Why traditional physics based models still remain important in forecasting

07:00 The training data AI models rely on

7:50 The uses of AI weather models and the advantages of being computationally light

09:16 Turing’s partnership with the MET Office – why is the weather important?

11:50 Introducing Fastnet: How cutting-edge weather models can tackle climate insecurities

12:30 How AI weather models differ from LLMs

13:15 How Fastnet uses an  icosahedral mesh to map the Earth.

16:05 Grey Swan events and Super El Niño

19:40Comparators to Fastnet and introducing a regional model for the uk

20:30 Improving the accuracy of fastnet

24:55 Green AI

26:30 The importance of diversity in AI modelling

28:30 Why it is important for the UK to have its own weather model

30:30 The future of Fastnet

32:13 Outro and find out more

This special episode of the Turing Podcast revisits Project Bluebird – the project that featured on our very first ever episode on the podcast in 2020! Host Amelia Jabry is joined by Ben Carvell, an aviation researcher leading AI agent design on Project Bluebird, and Andrew “Andy” Pace, a London terminal air traffic control officer (ATCO), who describes ATCO work, sectorisation, radar-based operations, and the still-analogue reliance on voice. They examine the growing complexity of air traffic control as traffic returns to, and surpasses, pre-COVID levels and the added challenges from drones, new aircraft types, and emissions constraints, stressing human cognitive limits in a safety-critical environment. Bluebird is a multi-year partnership between the Alan Turing Institute, NATS, and the University of Exeter with three themes: building a high-fidelity digital twin, developing virtual controller agents using methods from rules to optimisation and reinforcement learning, and addressing trust/explainability. Join Amelia, Andy and Ben as they discuss the challenges and opportunities of building an AI system to assist ATCOs in one of the most safety-critical industries in the world. 

 

00:00 The Hidden Chessboard at the heart of Air Traffic Control

00:37 Meet Ben Carvell and Andy Pace

03:20 What Air Traffic Control Officers (ATCOs) Do

06:20 Safety Rules for aviation and the moral responsibility of ATCOs

08:00 Introducing Project Bluebird

08:52 3 Main themes of Bluebird’s research: Digital Twins, AI Agents for Air Traffic Control, and Trust

10:30 The legal and moral responsibility of Air Traffic Control

12:20 The Science: The Physics Informed ML (PML) behind Bluebird. Including dealing with variability and the unknown.

14:55 Rules Optimisation versus Reinforcement Learning (RL)

15:50 A blast from the past

17:40 A collaboration with Johns Hopkins University tackling the issue of explainability in RL introducing ‘Relevant traffic’ 

20:36 TRUST: Safety Minima vs Collisions in Airspace – what is ‘non-negotiable’?

23:40 Interrogating the AI’s logic and maintaining trust

24:00 A curriculum for machines? The AI’s and ATCOs final exam

26:00 Bringing an ATCO onto project Bluebird – tool support for ATCOs

30:00 Intuition vs learning – the ‘mystical ATCO scan’ 

32:30 The ever-crowding airspace and increasing challenges to ATCOs

33:30 They history of Bluebird: 2020 versus now

35:40 How ATCOs really feel about Bluebird

36:02 Introducing Falcon - Are humans the ultimate guarantor of safety, or is our cognitive limit the ceiling for system efficiency?

40:15 Move 37 and the potential for Discovery

43:20 Collaboration between institutions and the real outputs Bluebird has made (the game!)

44:22 Wrap Up and Find Out More

How can the study of ant colonies and evolutionary theory help us ensure the resilience and sustainability of global shipping?

In this episode, host Amelia Jabry is joined by Professor Adam Sobey (Mission Director for Sustainability at the Alan Turing Institute) and Senior Applied Scientist Dr. Przemyslaw (Slaw) Grudniewski from Theyr. Together, they explore the ‘evolution’ of shipping route optimisation - from early concepts proposed by Alan Turing to cutting-edge Multi-Objective Genetic Algorithms.

Discover how these ‘survival of the fittest’ models are being used to navigate the complex world of charter party agreements, fuel efficiency, and autonomous vessels. They also dive into the environmental impact of rerouting, discussing how a 1% change in fuel consumption can protect vital megafauna like whales, and what the melting Arctic means for the future of global trade.

 

Chapter Markers

0:30 | Co-Host Introduction: Professor Adam Sobey

• Introduction of Adam Sobey, Mission Director for Sustainability at the Alan Turing Institute and Professor at the University of Southampton.

1:20 | Guest Introduction: Dr. Przemyslaw (Slaw) Grudniewski

• Introduction of Slava, Senior Applied Scientist atTheyr
• The history of Adam and Slava’s collaboration, starting from Slava's PhD in 2015.

4:30 | The Path to Genetic Algorithms

• Why the team focused on genetic algorithms, including the influence of a talk at the University of Bristol on co-evolution mechanisms.

5:00 | Why Shipping Matters: The Ever Given Incident

• The significance of global trade by sea (80-90%) and the 2021 Suez Canal blockage by theEver Given.

6:20 | The Sustainability Imperative

• Shipping currently accounts for 2-3% of world emissions, emphasizing the massive need for reduced costs and improved sustainability.

7:15 | Defining Genetic Algorithms

• Explaining unsupervised learning algorithms based on "survival of the fittest" and evolutionary mechanics.

8:40 | Applying Evolutionary Principles to Route Optimisation

• How routes are treated as individuals that create "offspring" through crossover and mutation.

10:20 | Multi-Objective Genetic Algorithms

• "There is no one best route"—balancing conflicting goals like voyage time vs. fuel consumption.
• Explaining why multi-objective approaches provide a set of optimal solutions rather than a single answer.

11:00 | Charter Party Agreements & Alternative Fuels

• The complexity of "rental agreements" (charter parties) and the shift toward net-zero fuels like ammonia, hydrogen, and nuclear.

12:20 | The Rise of Fully Autonomous Vessels

• Navigating the challenges of crewless ships and how they allow for real-time route adjustments.

13:30 | Sustainability Benefits of Autonomy

• Removing crew-related weight can lead to estimated fuel reductions of around 20%.

14:40 | Safety and Regulation

• The role of the Alan Turing Institute and Lloyd’s Register in developing standards and validation for autonomous systems.

16:15 | Risks: Cyber Threats and Bad Actors

• Addressing piracy, cybersecurity risks, and the safety of alternative fuel sources.

18:00 | Why Genetic Algorithms Win

• Comparing genetic algorithms against local search methods like A* and Dijkstra for complex, real-world problems.

19:00 | Top Performance: cMLSGA

• ThecMLSGA (Convolutional Multi-Level Selection Genetic Algorithm) and its 7-8% improvement over other models.
• This represents a saving of 50 to 380 tonnes of fuel per day for large vessels.

20:20 | History: From Alan Turing to Today

• How the field traces back to Turing’s 1948 ideas of "child-like" intelligence that learns and evolves.

22:20 | Ants, Tribes, and Co-Evolution

• Using the study of ant colonies and human tribal behaviour to understand collective fitness and reproduction.

23:50 | Scaling Solutions through Collectives

• Applying the concept of "collectives" to solve large-scale optimisation problems through collaboration.

25:25 | Multi-Level Selection

• How "groups of individuals" (collectives) can compete and work together to look at different objectives simultaneously.

26:20 | Collective vs. Convergence-Based Algorithms

• Why maintaining diversity in a population is more effective than focusing on a single "perfect" solution too early.
• Diversity provides better and more informed choices with the data at hand.

28:00 | Success Stories: TVOS and Whale Protection

• TheTVOS (Theyr Voyage Optimisation Software) and its real-world impact.
• The importance of protecting marine life and megafauna.

29:50 | Navigating the Environment

• The difficulty of rerouting massive ships and the role of deep learning vs. genetic algorithms in icy environments.

32:50 | The Arctic and New Trade Routes

• How melting sea ice is opening new routes and the resulting need for specialized "ice-class" vessels.

35:50 | Industry Adoption and Client Surprises

• Overcoming the lack of maritime background to deliver results that surprise industry veterans.

38:30 | The Future: Power Prediction Models

• What’s next for the field, including more advanced predictive modelling.

41:00 | Summary and the Power of Collaboration

• A hopeful look at how the Alan Turing Institute acts as a convening power for sustainability research.

The Arctic is undergoing rapid changes due to climate change, making sea ice forecasting increasingly important. In this episode of the Turing Podcast, hosts Amelia Jabry and Dr. Sophie Arana discuss the critical role of AI models like IceNet in predicting sea ice conditions and aiding conservation efforts. Featuring Dr. Ellie Bowler from the British Antarctic Survey, the conversation dives into the technical details of IceNet, its applications for wildlife conservation, and the challenges of predicting sea ice dynamics. The discussion also covers the importance of these predictions for indigenous communities and wildlife that rely on the frozen Arctic landscape. Learn how AI is revolutionising our approach to these urgent environmental challenges.

Read more about our environmental forecasting work here: https://www.turing.ac.uk/blog/democratising-environmental-forecasting-age-ai 

Explore IceNet: https://icenet.ai/

Find out more about Dr Ellie Bowler's publications: https://www.bas.ac.uk/profile/eller/

Read more about Dr Sophie Arana here: https://www.turing.ac.uk/people/dr-sophie-arana 

 

Chapters:

00:00 Introduction to Arctic Climate Change

00:39 Meet the your host Amelia and co-host Dr Sophie Arana 

00:57 Understanding the Role of a Research Application Manager

01:47 Introduction to ICE Net

02:47 AI vs. Traditional Physics-based Forecasting Models

03:40 Human Expertise and AI Collaboration

04:56 Introducing Dr. Ellie Bowler and her sea ice and caribou migration forecasting research

06:21 Challenges in Sea Ice Forecasting

10:01 Caribou Migration and Conservation

12:30 Impact of Human Activities on Arctic Wildlife

16:38 Innovative Conservation Methods

23:25 Future of ICE Net and AI in Conservation

26:54 Conclusion and Further Resources

New podcast - defending our critical national infrastructure.

 

Podcast host Amelia Jabry is joined by co-host and Turing Principal Research Scientist, Vasilios Mavroudis to explore how AI is reshaping both the defence and attack of vital systems like healthcare, energy and transport. With the help of special guest, Microsoft’s Director of Global AI & Cybersecurity Policy Nicholas Butts, they ask: how vulnerable are our power grids, water systems, and communications networks in the age of advanced AI? And what are organisations like the Alan Turing Institute and Microsoft are doing to protect these essential services?

Episode Notes

Please find below a list of resources and news articles that were mentioned in the podcast: 

• Microsoft Digital Defense Report 2024
• Glossary - AIRC
• Staying ahead of threat actors in the age of AI | Microsoft Security Blog
• https://www.thestack.technology/ai-agent-whisperer-liberates-llm-to-spout-filthy-cardy-b-lyrics/
• https://www.bbc.co.uk/news/articles/c0el31nqnpvo
• https://www.bbc.co.uk/news/articles/czx1qp64wrno

 

 

Ed and David chat with Professor Ibrahim Habli, Research Director at the Centre for Assuring Autonomy in the University of York, and director of the UKRI Centre for Doctoral Training in Safe AI Systems. The conversation covers the topic of defining and contextualising AI safety and risk, given existence of existing safety practices from other industries. Ibrahim has collaborated with The Alan Turing Institute on the "Trustworthy and Ethical Assurance platform", or "TEA" for short, an open-source tool for developing and communicating structured assurance arguments to show how data science and AI tech adheres to ethical principles.

On this episode, Bea and Jessie are joined by Becky Inkster and Mark Sheppard, co-founders of The Laß; a fusion of science, creativity and personal reflection.

Collaborating with famous musicians, dancers and members of the public, Becky and Mark use EEG brain activity technology to visualise poignant moments. They hope to offer both internal and external reflection opportunities to nurture personal growth and community bonds.

For more information visit -

https://www.beckyinkster.com/the-lab

 

Don't forget to explore our shorter AI news podcast - too long didn't read

To what extent can a computer network be actively managed and defended by intelligent autonomous agents?

In this episode, Ed and Anneca talk to Vasilios Mavroudis and Chris Hicks explore this question and more. 

Vas and Chris lead the Turing’s AI for Cyber Defence (AICD) research centre which seeks to fundamentally transform the way in which we secure digital systems through the development and application of cutting edge, deep-learning based approaches to intelligent agents. 

 

Find more information about AICD here

 

Don't forget to explore our shorter AI in the news podcast - too long didn't read

On this episode of The Turing Podcast Bea and Anneca are joined by Lord Chris Holmes, Britain’s most successful Paralympic swimmer and an active member of the House of Lords with a policy focus on digital technology for public good.

Connect with Lord Holmes on  Linked In

 

Explore our regular news AI news podcast, too long didn't read

The Turing Podcast revisits Project Bluebird; a fascinating collaboration aiming to solve some of the biggest and most complex problems in air traffic control with digital twins and AI.

Join Ed as he sits down with Nick Pepper of The Alan Turing Institute, George De Ath of the University of Exeter and Marc Thomas of NATS - the team behind Project Bluebird.

First featured on our podcast in 2020, the team now provides a progress update at the midpoint mark. Learn how they are developing innovative AI to train a digital twin air traffic controller with the aim of enhancing aviation safety and functionality, and what the challenges are, integrating human expertise with machine intelligence.