We recently caught up with our headline speaker Jeremy Wyatt Professor for AI & Robotics from the University Of Birmingham, Jeremy talks about his career researching AI & Robotics, what to consider when integrating AI & how brands can benefit from the technology.
You can hear Jeremy alongside a world-class line up of Digital leaders from BT, Instagram, Deezer, Met Office, Mobile Marketing Magazine, Pinterest + many more, don’t forget today is the last chance to bag yourself 2-4-1 tickets click this link to book – 241 Tickets
Jeremy Wyatt Speaker Interview Digital Enterprise Festival Birmingham
DEF – Welcome to our event Jeremy it is great to have you onboard, can you tell us about your career to date?
JW – Well, right now I’ve been working in robotics and artificial intelligence (AI) for 25 years. This means I’ve seen quite some changes in the field over time. In a previous life I was an arts graduate, then I moved into AI with an MSc at Sussex, then a PhD in machine learning at Edinburgh, followed by 20 years at Birmingham working on AI and robotics. My specialism is the combination of the two. I work on quite a number of different topics, including machine learning, machine vision, and probabilistic AI. My main expertise is in get robots to plan tasks in worlds that are new to them, and in robot manipulation. We tackle these using a combination of logic, probability theory, and machine learning. I’ve led a couple of large (20 person plus) research projects on these topics that led to two robots: Dora the explorer, and Boris. You can find videos of them on YouTube.
DEF – You are involved with a technology that is providing the biggest changes to business for decades, for those that are not AI experts please summarise what AI is & how it can help companies, departments & society?
JW – There are many types of AI, and AI has actually been providing value to businesses since the 1970s. The excitement—and I should also say hype—around AI at the moment is around machine learning. This is because machine learning allows you to turn data into decisions, and because the performance of the state of the art machine learning algorithms has improved steadily, so that we are now at the point where machine learning can outperform humans on some instances of basic perceptual tasks, such as some kinds of speech recognition or object recognition. The neat thing about machine learning is that it’s a generic technology, the algorithm is the same regardless of the task. But progress is still steady, not exponential, so businesses need to be careful. Also, curating data that is suitable for machine learning, and picking the right kind of machine learning for your problem, is a non-trivial exercise. Given these caveats, machine learning, AI more broadly, and also robotics can help businesses that make data based decisions make better ones, faster. In societal terms AI and changes in robotics are our major hope for improving flat-lining productivity and for filling what I call the demographic gap. This is the problem that because people are living longer our economies can’t grow quickly enough to support them, the only way forward is to become much more productive. I do think, however, that it is important not to overhype AI. We know it has potential, and we know that many companies already use it to good effect, but there are other disruptions coming that may be equally significant, and we are already living through an age where there have been a number of recent major disruptions.
DEF – Your audience will be made up of senior level executives from the world’s biggest brands, in your opinion what‘s the single biggest change marketers will see as a result of AI in the next five to 10 years?
JW – I think the answer in marketing is that AI may allow you to achieve several things. First, it will help to better understand your customers from the already available data on them—which is often too complex for humans to understand. Second, it will help you take decisions about messages that are personalised, or at least tailored differently to different segments. Third, it will enable you, ultimately, to run many different marketing experiments, each of which gives you further understanding.
DEF – A lot has been written about AI/Robots replacing people, would you agree the successful marketing teams in the future will combine machine & humans to achieve better results?
JW – I can’t speak for marketing specifically, but let’s take a similar market in advertising. It’s much more effective to have algorithms make ad buying decisions based on data. So that function should be outsourced to a machine, and checked by a human after. But for creative functions I can’t see a way to replace people. I’m also somewhat sceptical about certain advances that are touted in customer interaction, such as chat-bots. Autonomous dialogue systems are acceptable in constrained domains, but they are not up some of the jobs they are being asked to do. Also, I’m generally sceptical about job losses. The labour market continues to grow almost inexorably. The real issues are around de-skilling jobs, re-skilling humans whose jobs change, and sharing productivity gains fairly between capital and labour.
DEF – Which marketing skills will become more important, and which skills will become obsolete?
JW – Think about all the jobs the marketing department requires that involve data based decision making. Those are candidates for automation. Everything else is for humans.
DEF – In keeping with when new technologies come onto the market & make a big impact so called “experts” appear from nowhere … given your decades of experience in AI, Robotics & Machine learning research what important factors should companies take onboard before partnering new suppliers in this market?
JW – Do they have strong scientific and engineering heritage behind them? i.e. did they study at the best labs, if they published do they have publications in the best places? Can they tell you what underpinning technologies they are using and why? Also, frankly, if you are a non-technical person, you need someone who can help you interpret the answers. Realistically that means finding AI experts who’ve been around a while. I’ve noticed the need for this, and the phone rings far more often than it used to with requests for consultancy and expert advice. I’ve seen a roughly ten-fold increase in enquiries in the past year.
DEF – You have your own robots Dora & Boris, what is the story behind them & how do you decide on their gender/names?
JW – We pick gender names at random. My postdoc named Dora, I think because he had kids, so he knew the cartoon, and the robot explores, so it fits. I named Boris, just because I like the name. The story behind the robots is that Dora was an attempt to bring together many different strands of AI after 40 years of disparate work. Dora is dropped in to an unfamiliar world, and then asked to perform tasks. So imagine arriving at a new job one day, and being told to make the boss a cup of coffee. You wouldn’t know where anything is, so you find out the necessary information as you go along, e.g. search for room that looks like a kitchen, or ask someone. Dora has to solve these problems, and to do so the robot has to have an understanding of what it does and doesn’t know, and how the actions it can take can change not only the world, but it’s knowledge of the world. Dora is also one of the first robots that can explain her own failures. But, she has one limitation, which is that she can’t manipulate objects. Boris is a robot designed specifically for manipulating objects, and he uses machine learning and soft robotics to grasp and manipulate objects he’s never seen before. So, we’re trying to solve long-term problems in AI and robotics that are still, even today, largely unsolved.
DEF – Thank you for your time & insights Jeremy we look forward to your sessions on 10th November.
To view all of our fantastic speakers & their sessions click the links above or click here – AGENDA