Tell us about your job

I’m the head of data science and analytics at Digitas UK. When I am describing this to people who are not computer literate or particularly interested in data, I describe it as ‘understanding what people do and why people do it and using data and computers to predict what they’re going to do next.’  

In this role I oversee a few different kinds of teams, including data scientists and coders. We also have teams of analysts, so there are a lot of specialists. 

Public perception of data work can be varied, and people are concerned about privacy and data ownership, does this come into your role? 

There are different groups of people. Some are not bothered; they feel that they have nothing to hide and it’s just a way for data to be useful. Sometimes these algorithms really do make life easier. I like the way that Netflix learns my tastes and makes using it simpler for me. 

But I also think there’s a broad change in public mindset, especially with documentaries and information available on what happens when data use goes wrong. People are starting to understand that it’s not all about convenience and it's not all good. At the same time, I don’t think people know remotely enough about when things go really wrong, because they don’t always understand that data is everywhere. People don’t realise the extent to which data science models and algorithms influence their lives.

This is the most important part of my role. Shedding light on where things can go wrong and what we can do to prevent that from a practitioner, regulatory and organisational perspective is key. Today, you can’t live your life without having an algorithm touch it in some way, unless you live completely off grid.

What’s your mission? 

I am very much involved in helping clients and their businesses become more successful. Part of my role revolves around using data for positive outcomes and working to prevent the harm possible with the use of data. The goal is to work completely around using data for good and stopping some of the harm it could do.

How would you describe your journey to this point? 

I went to university and through that, landed my first role in the industry. That was the path that I took, and I was lucky to be able to do that. Getting into this field, you have to be good with numbers and problem solving. You absolutely need training, but it doesn’t have to be based on tertiary education. You can learn through boot camps and a lot can be self-taught. For anyone who is thinking about this, there are many online courses and free ways to dabble in data and code to find something you really enjoy. 

You don’t have to go to university. It’s a great option for some, not everyone – it might not suit the way they learn, and it can be inhibiting from a cost perspective. I’m pleased it’s no longer seen as compulsory.

In my role today, I’m successful, I’ve broken a glass ceiling and achieved a lot, but things still sometimes feel difficult to navigate. There’s being a woman and constantly dealing with this self-doubt many of us experience, and there’s being a woman of colour, and everything that comes with that. Going back and forth between the two is tough. 

I think younger people are better at navigating this. They are saying that this is not ok. I think the generation before me were not yet at the stage where they could push back and point out what is unacceptable, there were only a few and they were incredibly brave.

I believe my organisation has responded really well to the events of the last two years in recognising that they could have done more sooner and actually stepping up to do more. We’re also working on this with our clients, showing them that to be ready for the future, this issue is something they have to credibly address. 

Have you worked on things that help organisations be better? 

One of the things we’ve done is created a tool that anonymises CVs. It strips out all the details that can trigger bias in the hiring process, such as where the candidate went to school, career breaks and their gender. We also take out gendered terms in our job descriptions so we attract more people. 

Many of our clients use AI and we always work with them to think about why, what it is, and about the outcome. Some people assume that there’s no harm in using AI in marketing and advertising, but there are situations, like targeted ads on social media, where you have to be cautious. 

On the gender side, if you consider how some platforms approve ads for men's health versus women’s health, it’s fundamentally different. They still won’t approve a lot of ads around period products, they won’t let you show blood. They’re very reactive around nudity, even with pregnant women because historical data has linked naked women and blood to pornography and abuse. But if you don’t normalise things around periods or getting a breast exam, there can be horrific knock-on effects. Big advertisers don’t have problems with this, but start-ups need lobbying to get their ads through. 

Another example is from 2016 when the first global beauty competition judged by artificial intelligence software, Beauty.AI was launched. Nearly all the winners were white because the AI had been trained on data that established lighter skin as a sign of beauty. It’s just marketing, but in both of these cases the outcomes and impacts are detrimental. This is why we work hard to make our clients aware of the possible outcomes and rigorously work with them to make sure they are making the right choices about technology. 

It’s not just about getting a bad reputation for not thinking these things through, this is a business decision, where getting it wrong can have a financial impact, or a legal risk. 

Does the perception of science need to be changed to attract and retain women? 

There are studies that suggest that by the time girls get to a certain age they’ve already started to be convinced that STEM is not for them. It becomes ingrained because we give little boys toolboxes and Lego and we give girls dolls, but this is where problem solving should be nurtured. All kids love these things but if we keep reinforcing a message that it’s not for girls, we perpetuate the problem and it’s no wonder girls get to higher education thinking they don’t want to do a STEM subject. 

There are some great programmes for kids and teens that can help get them interested in coding and data topics, such as Girls who Code and Goldie Blocks, which is targeted at really young girls. Then we need to make sure that teenagers see women in these roles so they can be inspired. 

What can be done to eliminate bias?

I believe three things need to happen. Firstly, it starts with data practitioners because we can identify the biases that exist in data. I think young people, too, care about the impact of their work a lot more. It’s a knock-on effect of not waiting to do harm in the world, but also knowing that if your output and data set is full of bias, it’s not a good model. 

Secondly, non-practitioners need to learn too. When they ask their data teams for something to do with AI, they might not understand the implications of what they’re asking for. Not everyone in the C-suite needs to be able to code or build something themselves, but they do need to understand the implications of what they are asking for and what they will be putting out into the world. 

Thirdly, government regulation. Technology moves so fast, and it’s hard to regulate but we can do more. The EU has developed some guidance and the UK is catching up, but it is everyone’s responsibility to eliminate bias in tech. No one person can do it alone, it can’t just be the government, or just practitioners or companies, everyone has to get on board to stop harmful things from happening.