As part of the 'Businesses after COVID-19' webinar series, UCL FinTech Society interviewed Didier Vila (Global Head of Data Science at QuantumBlack).
Mo Safayat covers key discussion points about QuantumBlack’s business mission, the adoption of technologies by businesses, AI biases and data privacy concerns in this interview transcript.
1. What is QuantumBlack?
QuantumBlack is an analytics arm of McKinsey. We create impact for our clients by improving processes. We grow people by giving training to our data scientists.
2. How would you define AI and how much value it can add?
AI is the buzzword. In a nutshell, AI is the process to understand the previous behaviour, information to structure them and make decisions. Humans are at the centre of decisions but AI helps them make better decisions. AI is a subsegment of augmented intelligence.
3. What is hindering the adoption of these technologies by businesses?
The pandemic is changing everyone’s behaviour. Just look at how many are using public transportation and using notes and coins to pay. Nothing can stop the digitalization of society and the use of AI will continue as there is no barrier. More businesses are accepting this.
4. How to address the issue of bias in AI where the training data used may have some inherent bias?
Here at our company, this is critical: we develop a lot of resources to tackle this situation. Once we receive the data, we need to measure the bias in the data, we understand the underlying structure to identify bias. There are many measurements of bias, which can be done at an individual level or amongst the population level. There is no silver bullet. It is very complicated to measure bias. More importantly, overall self awareness of data scientists will move us forward in this matter.
5. Do you think companies are taking the ethics of data seriously enough or is there anything that can be done?
Pandemic made everyone aware of this, more digitisation will make us aware of this question. We are trying to tackle this by hiring a diverse team of scientists who can think differently and spot different things.
6. What do you think of the talent gap?
Every school and university needs to train in AI and programming. Government has to do a lot in this aspect. We also train and provide guidance to students, we have a different initiative called ‘Girls First”. On the macro level, every actor needs to be active to address this. At the micro level, everyone needs to use books and search online as you can learn everything online including data science.
7. Do you think the British public is right to be too concerned with data privacy? Is this slowing down progress?
This is a sociological question. I would like to shift this to a data science question. Essentially, from the technical standpoint, it is important to be transparent and explainable to everyone in society. We invest a lot of energy and resources to support our client to make decisions that are easy to communicate and are transparent.
8. With the pandemic, governments and institutions are using data to fight the crisis, do you think institutions are just catching up to applicability of data in this way? Is this a lasting change? Or have they been using it previously?
Data will become critical to make decisions. Yes we are going to a world fully digitised and data driven. A recent McKinsey report noted four directions of travel. The first is to put our energy to understanding customer expectations then we need to keep mathematical models updated. We need to make sure every organisation has the structure and technology to address the amount of data that is emerging in the pandemic and finally organisations need to have a culture where work can be done remotely and takes a data driven approach.
9. What are your thoughts on joining QuantumBlack or a small startup or a big tech company? What is your advice and tips?
I want students to do what they like. Every kind of career path has advantages, you need to think carefully, engage with senior people who will help you.
QuantumBlack will give you opportunities to work on solving problems in a lot of industries. Usually, consultants tackle 6-7 projects in two years depending on duration, tackle and apply LaTeX methodologies. Everyone brings something to the table based on their background. McKinsey training will be provided and you will learn about leadership, communication, project, diversity, technical skills, R&D innovation, fairness transparency, ethics, explainability and you will have a complete spectrum of skills after 3-4 years. You will also help and work with data scientists.