Most people have at least one crossroads moment in their life - when the choice they make defines their personal or professional trajectory. For me, it was being awarded an internship at Intel, the first one ever through Purdue’s Co-Op Engineering program in 1990.
At first, I questioned whether the then little-known Intel, California, or this internship (on the Pentium processor) was the right choice for me. I just didn’t know if I had the right technical skills for the work, or if engineering was really my path. But I took a leap of faith because I didn’t want to waste the opportunity. That internship gave me a fantastic insight into the day to day work of engineers, including a chance to prove to myself that I could do engineering! It grew into a very successful 18-year career at Intel and a 25-year career in tech. At that moment, I could have easily said “engineering isn’t for me” had I not had the nudge in the right direction, the vote of confidence, that this practical experience provided. Nearly 30 years later, I returned to Purdue last week, for their Distinguished Engineering Lectures programme, and had a chance to talk about the incredible career journey I’ve been on - in the hopes of inspiring a new generation of engineers as to what is possible.
Sometimes that extra support is the difference between saying yes or saying no - between following a path in STEM, or doing something completely different. Whether it’s inspiring self-confidence, offering reassurance or providing a financial safety net, showing support and removing the barriers that prevent individuals achieving their full potential can have a powerful impact.
At DeepMind we want to build advanced AI to expand our knowledge and find answers to some of the fundamental questions facing society. It is an ambitious and long-term goal, and we will only achieve it if we can bring people together with different experiences, knowledge, backgrounds. This is a field where diversity is paramount, not only for the innovative work that diverse teams produce, but because it’s vital that we mitigate the risks of bias in the development of algorithms and applications. We need as many perspectives as possible to make sure the important questions are being asked when it matters.
DeepMind Scholarships to open the field of AI
The DeepMind scholarship programme is one way we seek to broaden participation in science and AI. It gives talented students from underrepresented backgrounds the support they need to study at leading universities, and connect with our researchers and engineers. Scholars get their Masters' fees paid in full, as well as guidance and support from personal DeepMind mentors.
This week we announced the renewal and expansion of our scholarship programme with the University College London. Four more DeepMind graduate scholarships for students wishing to pursue a master’s degree in the Department of Computer Science will be available for students starting courses in 2020–21. But UCL is just one example. We also work with numerous other universities, such as Oxford, Queen Mary University London, the University of Cambridge and NYU, to broaden participation in AI and computer science.
I’ve seen many examples of the impact that diverse perspectives can have in practice. Take Shaquille Momoh, one of our DeepMind scholars, who was inspired to research protein folding prediction while studying at UCL. Nearly every function our body performs—contracting muscles, sensing light, or turning food into energy—can be traced back to proteins and how they move and change. Predicting their structure is fundamental to understanding the body, as well as diagnosing and treating diseases believed to be caused by misfolded proteins. Shaquille had a specific motivation for studying protein folding. He wanted to better understand sickle-cell anaemia, a painful inherited condition much more prevalent in black communities – and for which there is no current cure.
To ensure the next generation of researchers reach their full potential, protecting and strengthening the research and teaching capacity of our academic institutions is vital too.
DeepMind partners with a range of world-leading universities with the aim of extending research excellence and teaching capacity. We’ve established academic chairs in machine learning at the University of Alberta, University College London, and the University of Cambridge, offering unrestricted funding for world-renowned researchers to freely pursue their academic interests. These chairs will be supported in their research and teaching efforts by PhDs students. And many of our researchers hold dual affiliations, allowing them to continue teaching or supervising students at Cambridge, Oxford, Imperial, MIT, McGill and elsewhere (you can access some of these courses on YouTube).
Investing in the ecosystem
Only by investing the right way across the ecosystem will we able to ensure the highest quality AI research that benefits everyone. It’s also why we partner with charities like Chess in Schools and Communities and In2Science, and have become founding partners of the Deep Learning Indaba in Africa, Khipu AI in South America, the Eastern European Machine Learning Summer School, the Southeast Asia Machine Learning School (SEAMLS), and the AI4Good Summer Lab in Canada.
And to research how the lack of diversity affects the development of AI – how companies work, what products get built, and who benefits – last month we announced a research fellowship with the Partnership on AI to explore the pervasive challenge of developing AI for the benefit of people and society.
History has shown us that hard problems are best solved with collective effort. Innovation happens when people with different experiences, knowledge, and backgrounds join together, break down boundaries, openly share ideas and collaborate for a common goal. Building advanced AI responsibly may be one of the hardest scientific challenges to solve. If our sector can provide the right support for researchers and foster an open, collaborative and diverse academic culture, the impact could be truly transformative.