Blog

In the previous two blogs in the series, we addressed what makes machine learning models trustworthy and how machine learning models fail. This blog is focused on how machines learn and one particular approach called meta-learning, which enables machine learning models to improve over time such that they can generalise to perform related but distinct tasks. We will look at how humans learn compared to machine learning, what meta-learning is, and how meta-learning can be done responsibly.

The integration of Artificial Intelligence (AI) and machine learning has already brought transformative opportunities across the rapidly evolving landscape of the insurance industry. As data generation has surged and competition intensified, insurers have recognised AI’s game-changing potential as a tool for analysing expansive volumes of data with unprecedented speed and accuracy. With data generation occurring at such high velocity, insurers turned to AI as a series of experiments to make sense of this complexity and respond to the ever-increasing external shocks in the insurance market. But with expanded adoption comes a swathe of new and pressing concerns, and with AI advancements showing no signs of slowing, the insurance industry needs to consider the technological solutions that will tackle both the problems of today and the challenges of tomorrow.

AI is everywhere. It has rapidly become an integral part of how we organise our lives, go about our jobs, and get from place to place. And yet, in our discussions with customers, we often encounter a wide range of misconceptions about what AI actually is. In this article, we aim to shed some light on what this technology is and how it works, but also on where AI began, who the pioneers were that paved the way, and where all of this is heading.

At the beginning of 2023, we announced the launch of the Aioi R&D Lab - Oxford, a joint venture between Mind Foundry and our partners Aioi Nissay Dowa Insurance (ANDI) and Aioi Nissay Dowa Europe (AND-E). One of the projects the Lab is currently working on is to solve a problem affecting countries with ageing populations, like Japan and the United Kingdom. 

Taking the leap from the world of academia to pursue a career within industry can be a daunting proposition. After years spent studying and gaining skills in a very focused area, going out into the world and applying that knowledge to a broad range of applications has a unique set of challenges. We interviewed members of the Mind Foundry, Google and Oxa teams, to understand their experience of making the leap from academia to industry, sharing their motivations, challenges, and the advice they would give to those making or considering a similar move.

It’s hardly a secret that data is the lifeblood of the insurance sector. Some experts predict there will be over 30 billion connected devices worldwide by 2025, and the resulting explosion of data, coupled with huge leaps forward in the capabilities of AI and machine learning, has already begun having an effect on the insurance industry. 

We have seen Lemonade process a claim in only 2 seconds, and companies like Roots Automation exploring the use of Generative AI to improve customer service. But despite these advancements, many insurers still need to improve the bottom line as they experience a deceleration in the release of prior-year loss reserves and watch underwriting performances deteriorate in a post-Covid world. Access to data and the availability of new technology are not the issues, so why are so many insurers fundamentally failing to leverage new technology to tap into the potential of their data?

Why do we expect more from machines than we do from each other? We lean on machines for so much, yet when it comes to judging their performance, we seem to have a double standard. A human can fail, time and time again, and still be forgiven. But it’s not enough for a machine to consistently outperform a human - it has to be nearly flawless. Why is this? And is this the right approach?

In our first blog in this four-part series, we discussed how Machine Learning (ML) models might be considered ‘trustworthy’, and we emphasised challenges in bias and fairness, the need for explainability & interpretability, and types of data drift.

This blog will focus on how ML models fail, the value of model observability, monitoring and governance, and most importantly, how we can prevent model failure.

Generative AI (GAI) has been firmly at the centre of discussion ever since it was popularised by the launch of ChatGPT in November 2022. Although the insurance sector, with its focus on cost, has generally reacted with interest and excitement to the advent of Generative AI and Large Language Models (LLMs), there are limitations to the technology that represent significant obstacles to industry-wide adoption.  

Last year, Mind Foundry worked closely with Professor Brent Mittelstadt (of the Oxford Internet Institute) to get a deeper understanding of the law, policy, and standards in development surrounding machine learning. The output of this work was an in-depth research paper highlighting the key aspects of machine learning governance. This 4-part blog series summarises some of the key points from that paper to give a flavour of the most pressing problems in AI governance and some of the solutions that Mind Foundry are developing to solve them. 

Generative AI has exploded into the public eye, with the conversation focusing on the opportunities and risks it presents. However, in the Defence sector, there are specific challenges and problems that this technology is simply not suited to solve.

There’s been a lot of talk about the development of large language models (LLMs) recently, with the loudest voices split between those who think it poses a significant risk to humanity, to those who believe it could be the crowning achievement of all humankind. Between those two extremes are numerous reasons for optimism and many valid concerns about AI safety and how to use LLMs responsibly across many different sectors. Much of this hinges on ‘transparency’, ‘interpretability’, and ‘explainability’.

We interviewed Professor Steve Roberts, Co-founder at Mind Foundry and Professor at the University of Oxford, and invited him to share with us how he explains the meanings of these three terms to his students.

In February 2023, we announced the launch of the Aioi R&D Lab - Oxford, a joint venture between Mind Foundry and our partners Aioi Nissay Dowa Insurance (ANDI) and Aioi Nissay Dowa Europe (AND-E).  This is the first of its kind research lab, so naturally you may be thinking… What is it exactly? In this article we’ll tell you all you need to know.

The past few months have shown us how quickly deep tech innovation, and especially advances in Artificial Intelligence (AI), can be developed and adopted at great scale. New AI tools have emerged at breakneck speed, opening up a range of opportunities to positively impact problem-solving, decision-making, and efficiency. 

One of the most prolific to date, ChatGPT, has shown just how quickly advanced technology can arrive in the mainstream and capture the imaginations of experts and laypeople alike. This tool is just the latest step in a long and continuing line of advances in using AI to generate text. Its capabilities, which span everything from explaining quantum physics to writing poems, have sparked debates about the opportunities it could create and the impact it could have on the future of work.

Artificial Intelligence is already being used to solve problems in the Defence sector. But if AI is only seen as an opportunity to practice innovation, then the development of critical applications won’t progress fast enough.

The International Day of Women and Girls in Science, supported by the UN, is a fantastic opportunity to acknowledge the inspirational contributions of women across the globe in science and technology, who continue to break down stereotypes and pioneer gender equality in a traditionally male-dominated industry. We have the chance to inspire and encourage girls and women of all ages, backgrounds and abilities to discover their potential within science and become leaders in an exciting and innovative field.

To celebrate this important day, we are proud to showcase some of the extraordinary women at Mind Foundry who are making significant contributions to science and technology as they share their journeys so far and the advice they would give to others. Including:

Poppy, a Lead Research Scientist at Mind Foundry, leading the Defence & Security delivery teams to develop innovative solutions to complex problems.

Hannah, the Chief of Staff to the CEO at Mind Foundry, where she interacts with all aspects of the business from partnership building, sales operations, marketing, and strategy, with an additional focus to drive alignment.

Alessandra, a Senior Research Scientist at Mind Foundry, with a PhD in probabilistic Machine Learning and on a mission to push the boundaries of AI by integrating software solutions and original research. 

Harriet, a Project Manager at Mind Foundry, where she is responsible for managing a portfolio of Machine Learning projects in the Insurance sector and Product Manager of The Mind Foundry Academy.

Change is hard. We're creatures of habit by nature, and despite our best efforts, we often struggle to avoid repeating entrenched behaviours that take up much of our time and energy. It’s altogether too easy to find yourself coasting along the same path with no signs of improvement on the horizon. However, it doesn’t have to be this way. Setting clearly defined goals, ones with simple metrics to measure completion and success, is one of the most effective ways to break the cycle and set yourself on the path to change. This idea forms the bedrock of our goals process at Mind Foundry, where our amazing team of scientists, engineers, and designers work tirelessly every day to push the boundaries of human-AI collaboration. In doing so, they prove that our value is not just in the technology we create but in the people that create it. To that end, we wanted to share the stories of 3 members of our team and how they used our SMART Goals framework to take on some exciting and inspiring challenges outside of work.  

Mind Foundry recently hosted 'Defining Ambitions: The Future of AI in Public Sector', an invaluable virtual learning exchange for the public sector. Delivered in partnership with GovNewsDirect, the session's objective was to explore the possibilities for AI innovation within public services. It is fair to say that many organisations are only just beginning to use AI and machine learning; although growing, the number of organisations that rely on it as a core part of their operations is still relatively small, but those that do are drastically outperforming their peers. So, what does the future of AI hold in public sector organisations?

(This piece was originally published as a guest post for The Quantum Insider)

Helping the fight against malaria

Mind Foundry co-Hosts HumBug 2022, as part of the ComparE challenge, bringing together the greatest minds to help solve one of the world’s most important problems. The goal of the project is to use low-cost smartphones to detect the characteristic acoustic signatures of mosquito flight tones, helping understand the presence of potential malaria vectors. The goal of the Mosquito sub-challenge at ComparE was to develop algorithms to detect mosquitoes within real-world acoustic field data.

Mind Foundry are thrilled to have been included in the ‘Ethical AI startup landscape’ research, mapped by researchers at the EAIGG (Ethical AI Governance Group), who have vetted nearly 150 companies working in Ethical AI across the globe. This important research by the EAIGG is being conducted to provide transparency on the ecosystem of companies working on ethical AI. 

Between 2006 and 2010, there were 1,096 registered electric vehicles (EVs) in the UK. Since 2010 that number has exploded, and today there are an estimated 400,000 EVs on UK roads, not to mention more than 750,000 plug-in hybrids (PHEVs). With new technology inevitably comes new obstacles and complications and as EV numbers continue to soar, the provision of charging points to keep up with demand across the country is essential. However, recent figures suggest a growing disparity between the number of EVs and the level of charge point infrastructure, as well as a “regional divide” within the distribution of this infrastructure. The situation calls for a different approach to how EV infrastructure is implemented to ensure that, as more people switch to EVs, everyone who needs to access a charging point can do so, regardless of any geographical or socio-economic factors. Mind Foundry’s work in this field aims to provide the solution to this growing problem at a time where the UK government has just committed £1.6 billion to expand the UK's charging network under their new Electric Vehicle Infrastructure Strategy.

(A version of this piece was originally published in techUK as part of #QuantumWeek2021.)

We’re delighted to announce that we won a prestigious CogX award in the “Best Innovation in Explainable AI” category last night.

In June 2019, the government launched a call for evidence on the proposed framework of a strategy around data usage in the UK, receiving over 100 responses. At the same time, the government undertook a programme of roundtables, workshops, and webinars, to ascertain the parameters, objectives, and evidence supporting what would become the National Data Strategy. 

It’s no longer enough to measure model accuracy. Ethics, explainability, and wider performance are key KPIs that must be considered as equally crucial priorities.

August 20th is World Mosquito Day. A day to raise awareness about the diseases carried by mosquitoes and highlight the scientific innovations that are emerging to help us reduce the suffering caused by the world’s second most deadly animal. 

I would be surprised to find anyone who works in the tech sector, especially if they’re working with data, who hasn’t seen a significant emphasis on ethical applications of AI, or “doing AI ethically” (even the Pope has got involved!). Conferences, research, blog posts, videos, thought-starters are all - quite rightly - honing in on arguably one of the most important considerations of the 21st century: how do we build AI to the benefit of humankind?

To some aspiring to answer this question, this might signify decades’ worth of research. To others, it’s millions of hours of person-time in algorithmic design or troubleshooting software. The responsibilities to getting this right extend beyond this to policy, regulation, education, investment… the list goes on.

But the list isn’t the only thing that goes on; as I’m writing this, thousands to millions of companies around the world are grappling with adopting AI right at this very moment. They don’t have decades or even years to play with… they need it now. As I mentioned in a previous blog post, there’s a race on to get the most out of AI adoption before it’s too late. Currently, in the UK alone there are over 1400 high-growth AI startups and scale-ups, and this doesn’t even count the vast swathes of commercial and public sector adopters outside of the AI industry. This is a real challenge.

So how can you do it ethically? Or responsibly? Or is that even technically possible right now? Let me answer by addressing some of the most common questions that we’re posed at Mind Foundry.

Though the numbers fluctuate slightly year to year, only about 10% of the engineering workforce in the UK are female.

In the race to adopt AI, there is a flurry of activity happening in boardrooms and technical teams across the country. AI, which even a few years ago seemed to be the preserve of a vanguard of highly innovative companies, has suddenly become a prerequisite for organisations in every sector. Perhaps the stern warning from McKinsey’s 2019 report is ringing in their ears, that “Front-runners [...] could increase economic value by about 120 per cent by 2030” whereas “Laggards, who adopt AI late or not at all, could lose about 20 per cent of cash flow”.

It appears easy, then, to stay ahead of the curve and reap the financial benefits you need to adopt AI. Yet, according to MITSloan 2020 AI Global Executive Study, it’s not quite that simple, and only 10% of companies are obtaining significant financial benefits from AI technologies. 

So, why is that the case?

As AI capabilities advance, and with the advent of widely available low-cost cloud computing, AI will inevitably be applied to wider problem sets with more immediate and wide-ranging real-world impacts, bringing with it higher problem complexity and increased risk. As problem complexity and risks grow, the assurance of AI performance data cleanliness naturally transitions from pre-deployment considerations, instead becoming continuously evolving requirements which must be monitored and iterated on throughout a model’s life.

AI has the potential to help us tackle the problems associated with climate change and the warming of our Earth. The closer we get to the precipice, the greater the urgency. This has helped fuel tremendous growth in AI projects throughout government and the public sector, where AI is being used to make more accurate climate change predictions or to intelligently power the infrastructure that could support lower emissions on a global scale. 

Amidst all this enthusiasm, the one thing often being left out of the conversation is the carbon cost of these compute-intensive solutions. At best, the adoption of AI might be slowed down because people hadn't adequately considered the cost (financial or environmental) of the solution required. At worst, it could accelerate the warming of our planet. 

This is why it is so important to develop a Green AI technology: a technology that takes into account energy-efficiency as an important evaluation metric.

Decisions made by governments and other public sector organisations affect the lives of large numbers of people in profound ways every day. If considerations for ethics and responsibility are not made during the processes for designing, building, and implementing a solution with AI, unintended and unanticipated far-reaching consequences can be felt.

As software engineers we are called to make decisions. I would go as far as saying that most of our work revolves around it. From naming to choosing technologies, from picking the most appropriate conventions to outlining the best team process. Decision making is thus so deeply and subtly rooted in our craft that we often forget about its implications. What I want to explore today is the ability to settle for the lower end of the spectrum, i.e. choosing the simplest option and why you should give it more than a fleeting thought.

Machine Learning has become more and more popular. It's become accessible to a wider variety of users. And it is getting better and better at handling data - whether big or small. However one thing remains true about the data used in analysis - it is, most of the time, structured. Regardless of whether the data is stored in Excel spreadsheets, relational databases or big data repositories, it is structured.  That is to say,  it comes in the form of columns and rows of numbers, categories or labels. 

Today in so many industries, from manufacturing and life sciences to financial services and retail, we rely on algorithms to conduct large-scale machine learning analyses.

They are hugely useful for problem-solving and beneficial for augmenting human expertise within an organization. But they are now under the spotlight for many reasons – and regulation is on the horizon. Gartner projects that four of the G7 countries will establish dedicated associations to oversee artificial intelligence and ML design by 2023.

Many organizations still watch from the sidelines as competitors transform their understanding of data with AI. Why is this? 

By engaging with the right medical experts as early as possible in a drug’s lifecycle, pharmaceutical companies can gain enduring competitive edge. However, despite medical affairs departments continually undergoing profound change and growth, studies show 50% of new drug launches fail to meet company expectations. Could embracing AI in life sciences help to tackle this trend?

When patients suffer unintended reactions to medicines, it can be both dangerous for the individual and costly to society. However, what if medical professionals could use machine learning to forecast adverse drug reactions (ADRs) and minimise risks to patients?

For the past five years, data science has been praised as a technology that can unlock new applications and hidden insights for organisations. However, today it is struggling to live up to expectations.

Clinical research data plays an essential role in the pharmaceutical industry, but can also eat up resources in terms of time and money. However, by utilising machine learning in clinical trials, efficiency can be drastically improved, enabling faster and more cost-effective drug development and better patient outcomes.

The internet can seem full of scare stories and science fiction, but what is augmented intelligence really going to do to transform our lives in the near future?

When we consider the role of artificial intelligence in life sciences, it already has a proven track record as an effective tool to support pharmaceutical research. However, companies are also moving fast to explore the transformational impact it can have on their commercial operations, such as sales and marketing.

While machine learning has been around for decades, it has swiftly become a major buzzword in recent years. Following in the footsteps of 'digital transformation', many businesses are struggling to comprehend the intrinsic value of machine learning and the ways in which it can be implemented successfully.

Quantum computers have the potential to be exponentially faster than traditional computers, revolutionising the way we currently work. While we are still years away from general-purpose Quantum Computing, Bayesian Optimisation can help to stabilise quantum circuits for certain applications. This blog will summarise how Mind Foundry Optimise did just that.

Even with low cost widely available cloud computing, it can take significant time and compute power to train machine learning models on large data sets. This is expensive and is often at odds with the net-zero carbon goals of many organisations today. Throwing more data at a problem isn’t always the best answer, and by using AI that is responsible by design, we can reduce these problems while maintaining performance. Active learning is one of the methods we use at Mind Foundry to achieve this.

Hyper-parameter tuning can be observed in many everyday processes, but few people are aware of its existence. This blog aims to change that, and demonstrate its true value in data science.