The European Business Network
for Corporate Social Responsibility
Mobile Menu

Europe pushes for Ethical Artificial Intelligence

Friday, January 11, 2019

In its endeavour to prepare European societies for the digital future, the EU has made progress, but much is still to be done. Loubna Bouarfa is the CEO & Founder of OKRA Technologies and a member of the EU High-Level Expert Group on Artificial Intelligence. She provides recommendations to increase the use of AI in a more ethical way, such as in hiring employees.

According to Loubna Bouarfa, only 25% of European companies are adopting Artificial Intelligence, in sectors such as Fintech, healthcare, manufacturing and automotive. In this interview, she sheds light on the nature of Artificial Intelligence, how to address bias, and provides preliminary recommendations to increase the uptake of Artificial Intelligence in Europe.

 

1. What is Artificial Intelligence? How are successful companies using it nowadays? Any concrete examples?

Loubna Bouarfa (L.B): Artificial Intelligence (AI) is a set of systems or mathematical algorithms that display intelligent behaviour. They do so by analysing input from the real-world environment and detecting patterns in that data. In the same way that humans receive light and sounds through their eyes and ears and act accordingly, AI can receive signals from the real-world environment and take responsive actions with some degree of autonomy.

It is worth noting that there are many kinds of AI that are capable of learning, answering questions and providing recommendations. The most famous among them is the machine learning algorithm, which is my personal area of expertise. A machine learning algorithm is a framework or a formula for learning, designed by data scientists like myself, which allows the AI system to learn by itself. The role of this framework is simply to define the data input, the algorithm to be trained, which allows the machine to understand and adapt based on the information collected from the real world. This is the core of AI.

Each company uses Artificial Intelligence differently. This because AI adopts the values of each company in different ways and in different societies, depending on the goal it has been programmed to achieve and the specific data it uses. At OKRA Technologies, for instance, we use AI in the field of health and foster care.

·        In the field of healthcare, AI helps us to predict the best treatment for specific patients, identify the main predictors for a specific disease and to diagnose patients faster based on evidence from historical data of similar patients.

·        In the field of foster care, we use AI to efficiently match children with the most suitable foster families, by detecting early signs of a placement breakdown and predicting placement success, based on evidence from previous events experienced by other children.

This technology enables us to intervene quickly and effectively to achieve the best possible outcomes for human lives; as the AI system sends early warnings, it works to prevent illness and breakdown. In short, AI empowers us with tools to learn from the past events and maximize accuracy in predictions moving forward based on historical evidence.

2. People often talk about AI bias. What is this, and how can companies act responsibly in addressing it?

L.B: I think it is important to recognize that bias is ever–present in human decision-making; it can never be fully avoided, with or without involving AI. Bias is embedded in the human way of thinking. AI reads historical data, detects patterns, and predicts upcoming patterns, meaning that it replicates existing bias from past events. This can for example be problematic if you use AI in hiring employees and want to deliver on diversity hiring goals, because you might end up hiring the same kinds of people who are already doing well in the organization. However, the good thing about AI is that it enables us not only to identify bias, but also to measure the level of bias available in the data. This can be done in two steps:

·        First, we can balance and rebalance the data by feeding the AI system with several past events from different categories, instead of a single-event category.

·        Second, we can create a diverse workforce responsible for designing inclusive AI algorithms.

By doing so, we are tackling concerns with bias through a holistic approach, looking both at the data and the initial programming of AI systems.

With this holistic approach, the end-product goes beyond identifying bias, by making us act responsibly and counter-act bias. But this does not mean that we will have an entirely bias–free society. As long as there is uncertainty, there will always be bias. Artificial Intelligence can simply help us reducing bias as much as possible, but not entirely.

3. As AI solutions develop further, what are the main skills ordinary people should have to work alongside such intelligent systems?

L.B: Firstly, we need to recognize Artificial Intelligence as an augmenting rather than automating technology. When we think about AI, we often think of job losses and being made irrelevant, which distracts us from the potential of AI to save time and enable us to focus more on intellectual or creative tasks. AI can support us in harnessing our human skills such as empathy, our creativity and most importantly our emotional intelligence, and spending less time on things like administrative – related work. By leveraging our emotional intelligence, we will be able to work side by side with Artificial intelligence to improve our decision-making on data-driven evidence, reducing uncertainty, and ultimately maximizing our impact in the world.

4. Any recommendation?

L.B: I think Europe should move faster on the uptake of Artificial Intelligence. We are currently lagging behind the US and China. To this end, the largest European companies in sectors such as healthcare should foster collaboration with small companies who have more access to innovation and drive human outcomes at scale. Many of these small and medium enterprises (SMEs) have already embedded AI into their operational activities, and have a great ability to move fast, but their impact is restricted unless they collaborate with the largest companies. Such collaboration between large companies and SMEs is key to ensuring a greater and faster uptake of Artificial Intelligence in Europe.

 

Loubna Bouarfa was a speaker at the CSR Europe and Huawei conference on the ‘Digitalisation of work’ in Brussels last October 2018.