15 St Martins Schools students attended DeepMind’s facility in London, one of a range of activities undertaken by the school as part of its involvement in the Royal Society You and AI lecture series. Below is a summary written by three of the school’s students.
DeepMind is at the forefront of research into artificial intelligence (AI) and how its implementation can improve the world around us.
The first thing we noticed upon entering the building that housed the DeepMind HQ was how colourful, comfortable, and downright fun the office looked to be in; from soft and unique furniture to how almost every pane of glass looking into the central space had pixel art made using sticky notes. It was great to see so much diversity and inclusiveness amongst the employees and the relaxed and pleasant environment in which they work; all part of DeepMind’s company ethos.
We were given an intro to Deepmind itself and learned just how much of modern day technology uses AI and how much more will use AI in just 5 years time.
All activities put on by DeepMind engaged us and involved us participating events such as the User Interface (UI) creating challenge, to socialise with an unknown team (made up of students form the other three schools attending) and working together in a work environment.
The Robotics Lab Tour was one of the most interesting part of the trip as it allowed every student to question real engineers and coders who work for DeepMind on an AI project currently underway. During a fascinating lecture given by Piotr Mirowski, he explained how machine learning was developed and how machines can be taught from data, as well as how they can learn by themselves to complete tasks. We were fascinated about how seemingly simple tasks to us, computers would struggle with, such as differentiating between pictures of cats and dogs or working out how to play 2D computer games without being instructed beforehand. On the other hand, computers can get through so many calculations per minute, reviewing more data faster than humans ever can.
We were also given plenty of real world examples of AI and the use of it in modern technology. One of the examples that really stood out is how Google have used one of DeepMind’s AI’s on the energy systems that support one of their Data Centres, reducing its cooling bill by 40% helping them save money and most importantly resources, showing how they use the technology available to make systems more efficient, practical and environmentally friendly.
Throughout the day, every professional we met provided a real insight into how diversified and widely spread jobs in computing and AI are.
The instructors and computer scientists there presented their own journeys into DeepMind as an achievable and worthwhile goal for every student, and showed how coding was more than just a tool, it was a life skill that can be utilised in multiple careers, both practical and commercial.
One thing that particularly stuck with one of us was something that the main speaker of the day Pitor Mirowski quoted from Thomas J. Watson: “If you want to increase your success, double your failure rate”. This fits very well into machine learning as much as it does for a motivational speech on failure and success for humans; AI will fail hundreds of thousands of times at a task until they get it right, each time learning what it did right through either human supervision (supervised learning) or by something even as simple as a game score (reinforcement learning), or even sometimes through no feedback whatsoever (unsupervised learning).
The experience was educational and really showed us that AI will likely be one of the driving factors in industry in the future. It inspired us, and we are sure the others who attended, towards a career in the field and established the opportunities through which we can achieve our aspirations.