Diversity and inclusion in robotics: why women's voices matter

This article originally appeared in the July'23 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.

But even when a robot doesn’t possess any artificial intelligence (AI), biases can still be ingrained into it along the way – whether they are in the design, programming, or in the data that is used to train or programme it.

What is a robot and why do they need to be diverse?

Robots are physical machines designed to interact with the physical world. They can be autonomous or remotely controlled devices that perform tasks, manipulate objects, or carry out actions in the real world. Robots can be equipped with various sensors and actuators to perceive their surroundings and act accordingly. While some robots are programmed to follow pre-defined instructions, others can incorporate AI techniques to make decisions and adapt to changing situations. These physical machines interact with the real world, potentially incorporating AI techniques to enhance their capabilities.

If a workforce isn’t diverse, then training machines using potentially skewed data can lead to unintentional discrimination. As James Davis, Founder and CEO of inicio Health, attested to when speaking at the Decoding the Future of Women conference: “We tried training some AI on filtering out CVs, but we trained the model on all British names, and without realising it, it was filtering out those from other cohorts and demographics if the name didn’t sound culturally appropriate … it’s a terrible thing to do and it was completely by accident, but that’s exactly the issue it can cause.”

How do we shift the bias?

The obvious answer is diversity. But how do we diversify? And how do we know if we have got the balance right? As with most things, it is always a work in progress, and you don’t know what you don’t know until it’s questioned or challenged in some way.

Some potential ways to build diversity into robotics are:

• Look at your team: It’s 2023, and we know that diversity is crucial, but the statistics show us that diversity within teams is still hugely underrepresented when compared to the collective knowledge that it is not only important but necessary for growth. Building a diverse team is integral to incorporating multiple perspectives which, as long as the work culture supports differences of opinion, should collectively lead to eliminating unintended biases.
• Training using diverse data: Data should be collected from various sources, encompassing a whole range of demographics, cultures, and experiences – forgetting or discounting those who are underrepresented or marginalised will lead to unintended biases.
• Question evaluation criteria: Regularly evaluate the outputs of a robotics system to identify any bias behaviour. There should be metrics and evaluation criteria that establishes fairness and inclusivity.
• Use feedback for improvement: If someone is telling you that something is wrong, don’t dismiss it. This is valuable feedback and people with different backgrounds can provide crucial insights into the behaviour and impact of a system.
• Ensure ethical guidelines are in place: Establish ethical guidelines and standards for robot development. These must include items such as fairness, inclusivity, and diversity.
• Don’t sit on your laurels: There is always more to be done. Continually monitor, update, and improve the behaviour of robotics to ensure that any newly identified biases are addressed immediately, and then be willing to adapt and improve as new challenges and insights arise.

Aliya Grig, Founder and CEO of Evolwe AI, speaking at Decoding the Future of Women, comments: “What we put inside the AI, that will be the result that we achieve. Current AI models and robots are our kids, so they are learning from us, from our backgrounds, and our values.”

How are women helping?

A diversified group of engineers, researchers, and designers can bring a deeper understanding of the societal challenges that robotics and AI should aim to solve. If a company is using data from a limited pool of society to programme robotics, it can perpetuate and amplify any societal prejudice – so the chances are that if everyone has the same background there will be an unintended sway towards a particular inclination. By considering the perspectives and experiences of underrepresented communities, these technologies can be tailored to address specific social issues. It also ensures diversity is an integral component to building a machine, mitigating unintended discriminations, and thus promoting fairer representation.

Diversity also fuels innovation by uncovering new insights and solutions that may otherwise remain unexplored. Integrating these viewpoints helps to design robots and AI systems that can better address the needs and preferences of varied user groups, and it enables the development of more comprehensive and inclusive features – making the technology accessible to a wider audience.

As robotics continue to advance at an unprecedented pace, it is crucial to ensure that these  technologies reflect the different needs and perspectives of our society, and women are playing a pivotal role in shaping the future of diversity in robotics and are pushing the boundaries of what can be achieved in this rapidly evolving field.

Here are some ways in which women are actively involved in shaping the future of robotics:

• Research and innovation: Women researchers and innovators are pushing the boundaries of AI and robotics. Their work spans various areas, including computer vision, machine learning, natural language processing, human-robot interaction, and ethical considerations in AI. Their contributions lead to the development of more diverse, inclusive, and socially responsible technologies.
• Leadership and advocacy: Women hold influential positions in academia, industry, and organisations dedicated to AI and robotics. Through their leadership, they promote diverse perspectives, mentorship programmes, and initiatives that encourage women and underrepresented groups to pursue careers in these fields. They advocate for inclusive policies, equitable funding, and support systems that foster diverse talent and ideas.
• Education and outreach: Women in AI and robotics are actively engaged in educational initiatives and outreach programmes to inspire and empower the next generation of engineers. They participate in mentorship programmes, organise workshops, and speak at conferences and events. By sharing their experiences, they break down barriers, challenge stereotypes, and encourage more women to enter and succeed in AI and robotics.
• Ethical considerations: Women are highlighting the ethical implications and social impact of these technologies. They emphasise the need for responsible AI development, addressing biases, and ensuring fairness and accountability. Women are at the forefront of discussions around AI ethics, exploring issues such as algorithmic bias, privacy concerns, transparency, and the social consequences of automation.

Women in AI and robotics are designing the future of these fields through their research, leadership, advocacy, education, and ethical considerations. Their contributions not only enhance the technology itself, but they also foster a more inclusive, equitable, and socially conscious AI and robotic ecosystem. It is crucial to support and amplify the voices and achievements of women in these fields to ensure that the future of AI and robotics is as diverse, fair, and beneficial as it can be for everyone.