Shopping for an AI future at Ocado with the 'Data Whisperer' generation

Stuart Lauchlan Profile picture for user slauchlan December 4, 2017
A generation of 'Data Whisperers' is needed to exploit the full potential of the AI revolution, argues Ocado's CTO.

As we've covered before, members of the UK House of Lords Select Committee on AI are currently engaged in hearing evidence for a report on the “economic, ethical and social implications of advances in artificial intelligence”. This is due to be published in March next year.

Yesterday we looked at the recent expert testimony provided by a leading UK company, Dyson. Also representing UK industry was online grocery firm Ocado, whose Chief Technology Officer Paul Clarke described to the thirteen members of the committee how his company sees AI as one of its critical priorities:

Our business runs on an intersection of five disruptive technologies: the internet of things, big data, robotics, AI and cloud. Of those we see AI, in the Tolkien sense, as the One To Rule Them All. It is the one that lets you do the really exciting things with the others. It pervades what we do across our business, and yet, at the same time, like many others, we are just getting started here.

There are “amazing opportunities for consumers going forward”, he added, citing a number of examples to support his case:

There is faster lower-friction shopping, greater personalisation and adaptive user interfaces that can respond to different people’s shopping styles and agendas; better, smarter recommendations, predicting what customers want ultimately before they have a clue themselves. That is the journey we are on.

New kinds of interfaces are emerging—voice is a huge one at the moment—and augmented reality and AI are at the heart of those. There are increased levels of customer service on the back of greater reliability and efficiency. It is helping customers, and indeed employees, make fewer mistakes. For example, using AI for monitoring and oversight, to be a bit like the third gyro in the aircraft, sitting there on your shoulder trying to watch for when you are about to make a mistake, augmenting you as a human being.

It is about making better use of scarce resources, whether those be time, energy, water, land or transport networks. That is very central to what we do in the routing systems that we create: using all the data from today’s routes to drive smarter ones tomorrow. We are trying to keep ourselves safer, whether it be with physical or cyber security, and helping to manage our privacy. I am sure we will get to that later on. Those are very central to what we do at Ocado as an online retailer.

Building trust

When AI touches on the consumer experience, there’s a need for education and understanding in order to ensure mutual trust, said Clarke:

One of the reasons why it is so important that consumers understand the role of AI is that it is part of developing this trust between the providers of services and the consumer, both regarding the role that AI is playing and the role that the consumer’s data plays in enabling AI to provide those services.

What is particularly interesting here is that there is a slightly ‘Damned if you do, damned if you don’t’ situation. On the one hand, we know that our customers expect our systems to get to know them over time. They expect them to become knowledgeable about what they want. You could sum that up as someone saying, ‘I have been shopping with you for years. Surely you know I don’t like fish? Why do you still show it to me?’. On the other hand, consumers are equally capable of being offended if you make a conclusion about them, even if it is correct, such as, ‘How dare you assume I’m vegan?”. In a sense, the poor old AI is caught somewhere in the middle.

That is why trust and understanding is so important in order for people to realise that it is a new kind of contract, he added. That centres on data management:

Being clear about what you will do with people’s data is obviously very important and that includes, potentially, how it will be used to deliver smarter, better services on the back of that data. It is more than that, because we need to create new ways for this data to flow to more easily where appropriate. We have to create new kinds of data marts because open data is only a small part of what we will need.

We need ways for companies to be able to exchange data, but to do it with the appropriate kinds of passports and metadata that make it clear what that data can be used for. I am not talking here about personal private data, but other kinds of datasets that companies may want to exchange with each other. Some will have value but they will be very important, once again, if the UK is to make the most of this opportunity. If everybody keeps everything to themselves, we will not create the richness of the intersection of these datasets where some of the most exciting things are going to happen.

As Dyson’s Dr Mark Taylor also spoke about, the subject of available tech sklils, particularly around data science and AI, and the problems faced here:

Data science and AI are at the most overheated end of the computing skills spectrum. There is definitely a massive shortage…of graduates and post-graduates emerging with those skills. What is not talked about enough is the fact that those skills lie at the end of what is a pipeline of digital literacy that stretches all the way back to primary school.

There is a massive amount more that both government and business can do to help look after that pipeline across its entire length. Definitely, if we want a more diverse set and more graduates emerging at the end of the pipeline, we have to do much more at the start, whether it be, as the recent Royal Society report mentioned, making sure that we invest in more qualified teachers, or considering mandating schools to offer these subjects in digital literacy up to GCSE and A-level, and considering mandating these subjects in the curriculum up to GCSE like we do with maths and English.

That pipeline does not stop at university. It needs to continue on into industry, and we need to do much more to incentivise industry into that process of continual learning, particularly when it comes to subjects such as AI. There is no getting away from the fact that initiatives such as the apprenticeship levy have carved a hole in the available budget that companies, including ours, have to spend on that continual learning. We would argue very strongly that that might be better transformed into a training levy that can be used by companies in a much more holistic way to make sure that they continue that process.

We are going to have to fuzz this boundary between education and work in quite a considerable way, and we are going to have to see it as a much more hand-in-hand and continuous process of reskilling and reinvention across one’s lifetime. There is much more we could do there.

Clarke made an interesting distinction in his wording, talking about ‘digital literacy’ rather than ‘coding’. This is an important point, he insisted:

I see digital literacy as being a much bigger portfolio. It includes things such as how you harness data, how you visualise it, how you model it, how you understand bias. Data is core. It is one of the foods of AI. Teaching our children to have mastery of it, to be ‘Data Whisperers’ rather than just at the mercy of it, is extremely important. Understanding things such as the ethics and the philosophical challenges around these technologies is very important. We need to rethink the whole curriculum in terms of future-proofing it, because many of the things we are teaching our children now—and I say that as a father and as an employer—are going to be as disrupted as the encyclopaedia has been by the internet.

We need to develop people’s inter-personal skills, creative skills, intersectional thinking, strategic thinking, agile thinking—the process of reinvention. How do we teach our children to view reinventing themselves not as a mid-life crisis, but as a skill set, if you like, because they are going to do it many times? Preparing children in the widest possible way for this much smarter, more automated future is what we need to do, and we need to do it across the whole curriculum. Fundamentally, it is not just about teaching our kids to code. Viewing it like that is tinkering at the margins.

My take

What Clarke says about the importance of digital literacy is music to my ears. There’s often too much focus on coding as being the be-all-and-end-all of tech education. But the future isn’t just about technology; it’s about what we choose to do with emerging tech and how we allow/enable innovations to improve our lives and society as a whole. A generation of coders might be a useful asset to growing a healthy digital sector, but they need to be skilled up in ethical, moral and philosophical skills as well as being able to crunch code.

A grey colored placeholder image