Defined Crowd

Wanted: Women in AI

We live in a world where smart machines already dominate many areas of our lives. We use smartphones to order takeout, get directions, or book an Uber. And in the near future, advanced AI-driven innovations, such as facial recognition and self-driving cars, will become commonplace. 

But for a field that’s reshaping society in such profound ways, AI falls alarmingly short in its representation of women. As things currently stand, the people who work in AI don’t accurately reflect the society they wish to transform. This is problematic because it means that current AI tools may perpetuate existing structural inequalities.

In light of Women in History Month, we’re exploring the role of women in AI. We’ll focus not only on gender imbalance and bias in AI and ways to fix them, but also on some of the trailblazing women who are elevating the field to new heights, making history along the way.

Women in AI Stats

Gender imbalance is a serious problem in the field of AI. According to data from the World Economic Forum, women fill just 22% of global AI jobs, while men account for 78%.

Data also suggests that, although the field of AI professionals has grown substantially over the last four years, the percentage of women in AI remains low.

So, what exactly drives the gender imbalance in AI? For starters, the field has a distinct lack of prominent female role models. Without these, young girls may struggle to imagine themselves working in AI.

A 2018 Microsoft survey underscores just how important female role models are. They significantly increase the interest of girls in taking science and technology careers. That’s why it’s essential to showcase the achievements of women in AI far more prominently.

Another issue that discourages women is perceptions of the AI field as an unwelcoming and even hostile place. A male-dominated culture risks creating a toxic work environment in which women are perceived as inferior to their male counterparts.

What’s more, many girls still see AI as a ‘male only’ career path. Automated hiring processes are often biased, skewing towards male candidates. Facing such challenges, it’s unsurprising that women overlook AI as a potential career path.

Nevertheless, there are men in the field who actively support increased efforts to hire and retain great women.

Dhanaraj Thakur, Research Director at Center for Democracy & Technology, a non-profit based in Washington DC, says the key is to be intentional when seeking out women candidates. He told us: “We just have to look harder. That can mean reaching out to women’s networking groups, women in academia, and industry.”

Thakur also highlighted the issue of retention. He cited not only the importance of tackling harassment in the workplace, but also of implementing women-friendly HR policies, such as specific health coverage, maternity/paternity leave, and childcare—the latter especially key at industry events.

The Danger of Under-Representation

Gender bias in AI can be dangerous; namely due to the way the systems are trained. Without a substantial contribution from women, there’s a risk that machine learning tools will end up with biases baked in.

As Anima Anandkumar, a professor at the California Institute of Technology, points out, homogenous research teams increase the risks that AI systems will cause harm to certain groups. Diverse teams are not only better at spotting biased data, but also more likely to identify problems that risk negative effects on society.

Bias in AI systems often comes from the data itself. But what exactly does it look like when biased data is used in an AI system?

For example, collections of image data showing women doing domestic tasks like shopping, cleaning or laundry teach algorithms biased views of gender roles.

Another example is AI-driven risk assessment in criminal justice. Here, the algorithms could fail to acknowledge that women are less likely than men to reoffend, putting women at a disadvantage.

These are just two illustrations of the effects of gender bias in AI, among many. They highlight how critical it is for companies to use unbiased data to train machine learning algorithms. One easy way to make sure you’ve got unbiased data for your AI project is to use ethically collected and carefully vetted training data.

DefinedCrowd offers high-quality, prebuilt AI data sets, drawing on its proactively managed data collection process to combat gender bias in the datasets. That way, you can be confident that your AI project is making a positive contribution to society.

Addressing Gender Bias in AI

Gender bias in AI has become the focus of much attention in recent years. Companies have introduced measures to tackle gender bias in AI and move away from a male-dominated AI workforce. Let’s take a look at some examples of companies making inroads in this area.

  • Duolingo. The AI driven language learning app has achieved a 50:50 gender ratio in its engineering team. The company recruited from colleges with high numbers of women in computer science programs, reached out to women’s groups at those colleges and attended women’s networking events. Duolingo also gave its interviewers unconscious bias training.
  • Salesforce. Noted for its AI-driven Einstein Platform, Salesforce has spent $6 million to achieve equal pay for its women employees. The company conducts regular salary audits to maintain this situation and ensure an appealing environment for women in AI and tech.
  • Gusto. Cloud-based payroll software company Gusto decided to stop using typically masculine phrases in its job ads. Then it ramped up its recruitment efforts for female engineers. Now, over 24% of Gusto’s engineers are women, and the firm is committed to increasing that number.

Women in AI Take Action

According to LinkedIn research, AI specialist is the top emerging job for 2021. As a woman in AI reading this, what can you do to proactively empower your career? Let’s take a look at some of the best ways.

Strengthen Core AI Skills

One of the most important things you can do for your AI career is to improve your core skill set. The most in demand skills in AI and machine learning careers are the languages Python, R, and Java, along with TensorFlow and natural language processing (NLP). Numerous online programs teach these skills – be sure to pick a reputable one.

Plan Your Education Path

This depends on what point you’re at in your education. If you’re still choosing a specialism for your bachelor’s degree, the following subjects will be useful.

  • Computer Science
  • Information Technology
  • Mathematics and Statistics
  • Finance and Economics

If you’ve already completed your bachelor’s degree, consider taking an online training program from an accredited institution. Another good route is a master’s degree or PhD. These programs can be expensive, so it’s worth exploring funding options, especially for PhDs. 

Seek and Engage with a Mentor

No matter what stage you’re at in your AI career, having an inspiring mentor is an invaluable asset. Several initiatives already exist to match mentees with mentors for women in AI. Here are a few of our favorites:

  • WaiACCELERATE Mentors. Offers globally established experts passionate about responsible AI and entrepreneurship.
  • Women in AI Ethics. Trans-inclusive initiative from Lighthouse3 with a strong focus on AI ethics, that provides access to a network of global mentors.
  • SheCanCode Community. Offers career consultants and advice of the women in tech and AI, plus a job board with roles from companies dedicated to tackling gender bias.

Read About AI

Reading relevant books about the field gives added context and inspiration for your career. These can range from books about advances in the field, to more technical books about different aspects of AI technology. It’s useful to keep notes or write blog posts reflecting on what you’ve learned from different books. What’s more, sharing blog posts on social media is an excellent way to showcase your AI knowledge.

Pursue Internships in the AI Space

A career in AI demands experience as well as education. Doing an internship is a great way to get some experience. You’ll need solid basic skills, such as proficiency in Python, R or Java. Natural language processing experience is also useful, as are skills in linear algebra, statistics and calculus.

Build and Strengthen Your AI Community

Creating your own community is a great way to proactively shape your career in AI. You could launch your own AI event using a platform like meetup.com. If an AI group already exists in your location, why not create one specializing in a sub-area of AI, such as ethics or responsible AI. 

Become an Entrepreneur

One of the ultimate pathways to a career in AI is launching your own company. To be successful, scope the market and find places where AI can solve a problem for a large number of people. It should be a problem serious enough that people will pay money to solve it.

Influential Women in AI

Finally, here’s a quick introduction to some of the world’s most influential women in AI, including specialists in research, policy, ethics and entrepreneurship.

Women in AI Research

  • Daniela Rus @MIT_CSAIL

Director of MIT’s computer science and artificial intelligence lab. Daniela’s research interests include robotics, mobile computing and data science.

  • Sara Hooker @sarahookr

AI Resident at Google Brain, researching in deep learning. Sara is interested in security and algorithm interpretability and is a strong believer in AI ethics.

Women in AI Policy and Ethics

  • Meredith Whittaker @mer_edith

Research Scientist at NYU, founder of Open Research group at Google. Meredith has advised the White House, the European Parliament, and numerous governments on aspects of AI, internet policy, privacy and security.

  • Kay Firth-Butterfield @kayfbutterfield

Head of Artificial Intelligence and Machine Learning at the World Economic Forum. Kay helps governments around the world craft AI policy.

Women in AI Entrepreneurship

  • Dr Vivienne Ming @neuraltheory

Vivienne is the founder of Socos Labs, a firm that explores the future of human potential. She has also created AI tools to treat diabetes, predict episodes of mental illness, and help reunite refugee families.

  • Carol Reiley @robot_MD

Carol is the founder and CEO of AI Roboticist. She owns eight technical patents and has published numerous research papers. Carol’s research focuses on intelligent robotic systems to help humans with skilled tasks.

  • Daniela Braga @DanielaRBraga

Daniela Braga, Founder and CEO of DefinedCrowd, one of the fastest growing startups in the AI space. With over eighteen years’ experience in speech technology in both academia and industry, alongside her CEO role, she regularly guest-lectures at the University of Washington and has authored more than 90 scientific papers and several patents.

Daniela Braga, CEO of DefinedCrowd, in Lisbon

To discover more examples of brilliant women in AI, check out Lighthouse3’s 2020 list.

Key-take-a-ways

Any system that intends to change the world needs full participation from women. AI systems are no exception. Women are still under-represented in AI, but signs of change are emerging. Many companies now recognize how vital it is for their AI teams to include women, especially to reduce the likelihood of harmful biases entering AI systems.

Inspirational female leaders are making new inroads in the field, acting as powerful role models for the next generation of women in AI. Despite its many challenges, Carol Reiley of AI Roboticist is confident that the field is full of opportunities for women. She told us: “Don’t wait. Dive in and get your hands dirty. Define and create the job you want for yourself.”