I was recently joined by my Capita One colleague Simon Smith, Damilola Adewuyi, Head of Business Analysis at Government Digital Service, and an excellent group of participants for a vibrant discussion around predictive analytics. In the room were individuals from a variety of backgrounds, from government and housing associations to not-for-profit, consultancy and even a graduate hoping for a future as a data scientist. This fantastic mix led to a really interesting session and dynamic discussion.
Changing the future
After a brief introduction by Robin Knowles, CEO of Digital Leaders, we started the session by thinking about what predictive analytics is, and how it can be used to recognise signs that an event or situation is likely to occur, meaning interventions can be made to prevent the situation before it happens. For example, by understanding some of triggers for young people becoming NEET (Not in Employment, Education or Training) you can identify children showing the same early indicators and be able to take appropriate preventative measures.
Bots, bots everywhere!
Chat-bots are a simple example of predictive analytics, and they are ubiquitous on the internet these days. Some are so well trained that they would surpass the Turin Test by a country mile. However, we considered how the training of these bots cannot be underestimated, and the issue of transitioning from a bot to a human when the conversation is more complicated is yet to be solved in an elegant way.
Where do we draw the line?
When predicting outcomes for a person how safe is it to trust a machine? GDPR comes into force in May 2018 the rights for individuals to not have automated decisions made for them become much clearer, and consent will need to be obtained. It is therefore critical that we ensure this continually advancing technology is used in an ethical and safe way, and careful governance will need to be in place to offer individuals the right protection.
An artificial intelligence takeover?
Next up was Damilola Adewuyi, with his views on predictive analytics and machine learning. Damilola warned of a future where advancements in technology could go beyond predictive analytics into areas of AI and Deep Learning which could cause an ‘Orwellian’ end to humans. He considered how machine learning is currently used, and the amount of information technology is taking from people. And what risks does democratised (user-centred – i.e. free and open-source software) innovation within software and technology in the public sector pose for us? Damilola also considered the importance of considering user needs and not allowing the advancement of AI to go too far.
Lots of debate continued in the room and we spent some time discussing the legislation we have in place to govern and manage AI.
Deep data discussions
We agreed that predictive analytics is only as good as the data from which predictions are made. Data management is therefore key in avoiding poor data, which in turn produces poor predictions.
We also considered how analytics, particularly predictive analytics, should be a business function and not an IT function. We concluded that the questions which need to be asked of it should be business ones, and the technology employed should empower the business to answer them.
A big thank you from us at Capita to Robin Knowles, Damilola Adewuyi and the dynamic group of participants for joining our Digital Leaders Salon and contributing to a truly thought-provoking afternoon.