Organizations need diversity now more than ever, but it's the duty of everyone in business to ensure that people from different backgrounds feel confident and included.
That was the conclusion of a panel at the Mesh AI Data and AI Symposium in London, where IT professionals discussed the role of women in data and the benefits to be gained from a diverse workforce. The conversation took place within the context of her three issues: addressing bias, building careers, and attracting and retaining talent.
deal with prejudice
Clementine Whitcomb, a data engineer at EDF, said it's important to recognize that AI models are trained on real-world data, which can contain biases and biases. I insisted. Companies need to build teams that look beyond homogeneous experiences so they can check the model.
Having many different perspectives and experiences as part of the development process helps ensure those biases don't go unchecked. This is an important part of not blindly trusting what these algorithms give you.
Valeria Cortez Vaca Diez, Senior Data Scientist at Monzo Bank, talked about the challenges she faced when applying for a credit card a few years ago. When she started working at Monzo, she discovered that her application had been rejected because her full surname contained a compound not normally found in UK data patterns.
Cortes said her example highlights the problems companies face if they don't handle their data carefully. If you don't include different perspectives as part of your design process, you may miss the following issues:
With AI, it's not obvious where the biases are, and the lack of diverse experience means we can't detect where it might be causing harm. Therefore, brainstorming of different perspectives relevant to the client is not done.
When Cortez helps build models to fight financial crime at Monzo Bank, a key part of success is making sure humans ask the hard questions about what data goes into the models. That's it. We also look for well-known biases.
We will see what is included in the model. Don't just blindly pass everything through. There's a lot of his second validation to see how the model works. We focus on explainability and pay attention to the data people are choosing into their models.
building a career
Despite calls for diversity, the IT industry remains male-dominated, with some estimates suggesting that women make up around 26% of tech workers. Whitcomb said it was the people he has worked with at EDF that helped establish his career in the data profession.
It's about having a support network of role models who support and encourage me. What has helped me in my career is an open culture where people listen and collaborate. It's about the idea of ​​helping people make the best decisions and become their best selves. That's the key difference.
Ms. Whitcomb has benefited from working with senior women leaders in the business world. She says people need role models that reflect her own background, whether that experience involves ethnicity, education, sexuality or disability.
Diversity within a company creates a sense of belonging, knowing there is a community of people to share experiences with and seek advice from. Also, meeting a leader you can relate to can help you envision your future at the company.
Cortés began studying in Germany and found that he was often the only non-German in the room. Since she moved to the UK and worked at Monzo, she has become more confident.
In Britain it was different. Because I suddenly found myself in a much more diverse group in terms of the people I meet, especially in the technology field. Having a safe environment has made me feel more confident in myself and no longer shy about what I am capable of.
Representation is important and managers should strive to create a safe environment, she said. First, make sure the people in the room are from a variety of backgrounds. Second, ensure that the pursuit of diversity is an ongoing effort.
I think this approach creates a bit of a safe space when you know you're not “weird” just because you have a different background or experience.
Attract and retain talent
According to Whitcomb, it's important for organizations to recognize that attracting and retaining diverse talent are two different challenges. When attracting employees, companies must be careful to understand that unconscious bias can impact the hiring process.
Biases only become a problem if we are unaware of them and do not take steps to prevent them from influencing the decisions that are made. That means focusing on things like unconscious bias training, striving to have as diverse a committee as possible, and considering conducting blind resume screening. This means removing any information on a resume that could be used to categorize someone into demographics, and instead basing hiring decisions on that. their skills.
Cortes argued that the goal of data managers should be to ensure that diversity is always considered as part of the hiring process. The key to success is using data to track and follow your approach.
How are promotions changing across different groups? Do you want to be able to measure that to understand if something is going wrong or if something needs to be adjusted differently? I'm thinking of making it. Therefore, try to measure everything, see where things are not good and be proactive.
Cortes says it's important to not only consider quantitative results, but also focus on qualitative sentiment. If you want to retain your staff, ask them how they feel about the company's culture. It's important to foster conversation and ensure that people of all backgrounds are not ignored.
It's about having a culture where people feel like they can make suggestions to make them feel more comfortable or feel more part of a community. It's easy to forget that we're not holding a guessing game and can ask people about their experiences.
my view
Interesting discussion on a series of ongoing issues. For even more insight on this subject, check out Madeline Bennett's “Things I Wish I Could Have Said Back then…” series of articles.