Chris Bishop leads Microsoft’s research lab in Cambridge, which has been at the forefront of AI, machine learning and deep learning research for 20 years. Its work contributes to many Microsoft products and features, such as Clutter in Office.
Name: Chris Bishop
Role: Technical Fellow and Laboratory Director
Lives: Cambridge, UK
Family: Wife and two sons (both at university, studying Biology and Computer Science)
Pets: Two cats
Hobbies: Flying aeroplanes
Tell us about your current role?
I was one of the first people to join Microsoft’s Research Lab in Cambridge UK, back when the lab was first opened in 1997, before being named Lab Director two-and-a-half years ago, so I’ve been involved in growing and shaping the lab for more than two decades. Today my role includes leadership of the MSR Cambridge lab, as well as coordination of the broader Microsoft presence in Cambridge. I am fortunate in being supported by a very talented leadership team and a highly capable and motivated team of support staff.
What were your previous jobs?
My background is in theoretical physics. After graduating from Oxford, I did a PhD in quantum field theory at the University of Edinburgh, exploring some of the fundamental mathematics of matter, energy, and space-time. After my PhD I wanted to do something that would have potential for practical impact, so I joined the UK’s national fusion research lab to work on the theory of magnetically confined plasmas as part of a long-term goal to create unlimited clean energy. It was during this time that there were some breakthroughs in the field of neural networks. I was very inspired by the concept of machine intelligence, and the idea that computers could learn for themselves. Initially I started applying neural networks to problems in fusion research, and we became the first lab to use neural networks for real-time feedback control of a high-temperature fusion plasma.
In fact, I found neural networks so fascinating that, after about eight years working on fusion research, I took a rather radical step and switched fields into machine learning. I became a Professor at Aston University in Birmingham, where I set up a very successful research lab. Then I took a sabbatical and came to Cambridge for six months to run a major, international programme called “Neural Networks and Machine Learning” at the Isaac Newton Institute. The programme started on July 1, 1997, on the very same day that Microsoft announced it was opening a research lab in Cambridge, its first outside the US. I was approached by Microsoft to join the new lab, and have never looked back.
What are your aims at Microsoft?
My ambition is for the lab to have an impact on the real world at scale by tackling very hard research problems, and by leveraging the advantages and opportunities we have as part of Microsoft. I often say that I want the MSR Cambridge lab to be a critical asset for the company.
I’m also very passionate about diversity and inclusion, and we have introduced multiple initiatives over the last year to support this. We are seeing a lot of success in bringing more women into technical roles in the lab, across both engineering and research, and that’s very exciting to see.
What’s the hardest part of your job?
A core part of my job is to exercise judgment in situations where there is no clear right answer. For instance, in allocating limited resources I need to look at the risk, the level of investment, the potential for impact, and the timescale. At any one time there will be some things we are investing in that are quite long term but where the impact could be revolutionary, along with other things that have perhaps been researched for several years which are beginning to get real traction, all the way to things that have had real-world impact already. The hardest part of my job is to weigh up all these factors and make some difficult decisions on where to place our bets.
What’s the best part of your job?
The thing I enjoy most is the wonderful combination of technology and people. Those are two aspects I find equally fascinating, yet they offer totally different kinds of challenges. We, as a lab, are constantly thinking about technology, trends and opportunities, but also about the people, teams, leadership, staff development and recruitment, particularly in what has become a very competitive talent environment. The way these things come together is fascinating. There is never a dull day here.
What is a leader?
I think of leadership as facilitating and enabling, rather than directing. One of the things I give a lot of attention to is leadership development. We have a leadership team for the lab and we meet once a week for a couple of hours. I think about the activities of that team, but also about how we function together. It’s the diversity of the opinions of the team members that creates a value that’s greater than the sum of its parts. Leadership is about harnessing the capabilities of every person in the lab and allowing everyone to bring their best game to the table. I therefore see my role primarily as drawing out the best in others and empowering them to be successful.
What are you most proud of?
Last year I was elected a Fellow of the Royal Society, and that was an incredibly proud moment. There’s a famous book I got to sign, and you can flip back and see the signatures of Isaac Newton, Charles Darwin, Albert Einstein, and pretty much every scientist you’ve ever heard of. At the start of the book is the signature of King Charles II who granted the royal charter, so this book contains over three-and-a-half centuries of scientific history. That was a very humbling but thrilling moment.
Another thing I’m very proud of was the opportunity to give the Royal Institution Christmas Lectures. The Royal Institution was set up more than 200 years ago – Michael Faraday was one of the early directors – and around 14 Nobel prizes have been associated with the Institution, so there is a tremendous history there too. These days it’s most famous for the Christmas Lectures, which were started by Faraday. Ever since the 1960s these lectures have been broadcast on national television at Christmas, and I watched them as a child with my mum and dad. They were very inspirational for me and were one of the factors that led me to choose a career in science. About 10 years ago, I had the opportunity to give the lectures, which would have been inconceivable to me as a child. It was an extraordinary moment to walk into that famous iconic theatre, where Faraday lectured many times and where so many important scientific discoveries were first announced.
One Microsoft anecdote that relates to the lectures was that getting selected was quite a competitive process. It eventually came down to a shortlist of five people, and I was very keen to be chosen, especially as it was the first time in the 200 year history of the lectures that they were going to be on the subject of computer science. I was thinking about what I could do to get selected, so I wrote to Bill Gates, explained how important these lectures were and asked him whether, if I was selected, he would agree to join me as a guest in one of the lectures. Fortunately, he said yes, and so I was able to include this is my proposal to the Royal Institution. When I was ultimately selected, I held Bill to this promise, and interviewed him via satellite on live television during one of the lectures.
What inspires you?
I love the idea that through our intellectual drive and curiosity we can use technology to make the world a better place for millions of people. For example, the field of healthcare today largely takes a one-size-fits-all approach that reactively waits until patients become sick before responding, and which is increasingly associated with escalating costs that are becoming unsustainable. The power of digital technology offers the opportunity to create a data-driven approach to healthcare that is personalised, predictive and preventative, and which could significantly reduce costs while also improving health and wellbeing. I’ve made Healthcare AI one of the focal points of the Cambridge lab, and I find it inspiring that the combination of machine learning, together with Microsoft’s cloud, could help to bring about a much-needed transformation in healthcare.
What is your favourite Microsoft product?
A few years ago, the machine learning team here in Cambridge built a feature, in collaboration with the Exchange team, called Clutter. It sorts out the email you should pay attention to now, from the ones that can be left to, say, a Friday afternoon. I love it because it’s used by tens of millions of people, and it has some very beautiful research ideas at the heart of it – something called a hierarchical Bayesian machine learning model. This gives it a nice out-of-the-box experience, a sort of average that does OK for everybody, but as you engage with it, it personalises and learns your particular preferences of what constitutes urgent versus non-urgent email. The other reason I’m particularly fond of it is that when I became Lab Director, the volume of email in my inbox quadrupled. That occurred just as we were releasing the Clutter feature, so it arrived just in time to save me from being overwhelmed.
What was the first bit of technology that you were excited about?
When I was a child I was very excited about the Apollo moon landings. I was at an age where I could watch them live on television and knew enough to understand what an incredible achievement they were. Just think of that Saturn launch vehicle that’s 36 storeys high, weighs 3,000 tonnes, is burning 15 tonnes of fuel a second, and yet it’s unstable. So, it must be balanced, rather like balancing a broom on your finger, by pivoting those massive engines backwards and forwards on hydraulic rams in response to signals from gyroscopes at the top of the rocket. It’s that combination of extreme brute force with exquisite precision, along with dozens of other extraordinary yet critical innovations, that made the whole adventure just breath-taking. And the filtering algorithms used by the guidance system are an elegant application of Bayesian inference, so it turns out that machine learning is, literally, rocket science.