The world is waking up to the wealth of opportunities to leverage Artificial Intelligence (AI) to address the converging crises we face as a global community; from climate change, economic inequality and war to societal polarisation and growing distrust. While AI has enormous potential to bring people together, it can also drive misinformation, propaganda, hate and civil unrest. Now, as AI increasingly informs the decisions and processes that shape our lives, it is more critical than ever that AI solutions are rooted in humanity and inclusion.
As momentum around AI grows and 2022 comes to a close, The Women4AI Daring Circle remains committed to accelerating the uptake of inclusive AI and facilitating ecosystem transformation, through inspiring companies to act and pledge their commitment to inclusive AI whilst adopting effective policies to drive these commitments forward.
The Daring Circle defines inclusive AI as “AI which accounts for needs of different groups, including minorities, marginalised and underrepresented groups. The application of inclusive AI addresses the issue of bias and discrimination with the aim of reducing inequalities, including representation, accessibility and interpretability.”
At the 2022 Women’s Forum Global Meeting in November, the Women4AI Daring Circle launched a Call to Action which brings together signatory organisations to commit to implementing inclusive AI. It focuses on inclusion beyond representation, urging organisations of all sizes, from all sectors, to make concrete, process-based commitments that are reflective of the changing AI environment. Prominent organisations – including BCG, BNP Paribas, L’Oréal, Publicis, Standard Chartered and Clifford Chance – embraced the Call to Action and the opportunity to accelerate their journey’s toward inclusive AI.
Alongside the launch, Rim Tehraoui, CDO BNP Paribas and Genevieve Smith, UC Berkeley shared their insights with the audience around ‘How to make inclusive AI a reality’. Genevieve highlighted the urgent imperative to act: “People are interacting with AI every day: applying for jobs, loans, healthcare, social media algorithms. It’s crucial these are developed inclusively, before it’s too late. If you don’t prioritise inclusion, the default you’ll end up with is exclusion.” The cost of inaction is high – for individuals, communities and organisations alike.
As more businesses reap the benefits of AI, the onus to take action to ensure inclusion is increasingly falling on companies themselves. Last year, 56% of organisations surveyed in McKinsey’s Global Survey on AI had adopted Artificial Intelligence in at least one of their business functions – compared to 50% in the year prior – and nearly two-thirds said their companies’ investments in AI will continue to increase over the next three years. As AI grows in scope and power, it is more important than ever that it is designed with inclusion at its core, both in its development and its application. As Rim Tehraoui said at the Call to Action launch, “stereotypes and other forms of bias cannot be allowed to affect access to essential services. Algorithms need to be robust, explainable, and keep humans at the centre of Artificial Intelligence.”
The Call to Action’s first theme: ‘Strengthening pipeline & practices’
According to Fortune Business Insights, in 2021 the global market size for Artificial Intelligence was USD 328.34 billion – a quickly growing industry with potential for sustainable, inclusive growth and employment. Yet, the World Economic Forum found that only 22% of AI employees globally were women.
This gap in the AI market reflects the gender and skills gaps seen all too often in technology sectors. For organisations to implement inclusive AI, they must start addressing this imbalance internally.
Embedding the values of inclusive AI throughout an organisation requires the transformation of systems and processes, starting with how teams are built. Effective design and development teams should include a mix of disciplines and identities – gender, racial, religious, sexuality, nationality, disability, neurodiversity, generational, and others. Providing capacity building and support mechanisms can help drive diversity at organisational level, while ensuring that existing teams are trained to maximise the potential of inclusive AI.
The Call to Action’s second theme: ‘Build solid data foundations’
It is crucial that AI solutions use complete and representative data, to avoid the risk of replicating existing biases and exclusive practices. To build trust in AI solutions, robust systems to ensure that data is representative, inclusive and accurate must be in place.
For example, a large online retailer designed an AI software to identify the best job applicants, using a system that assigned scores of one to five stars using data from previously successful applicants. This approach replicated existing biases, thereby penalising applications that included words such as “women’s” (i.e. “women’s chess club captain”) and downgrading graduates from two all-women’s colleges. The system favoured words such as “executed” and “captured”, which were featured more often on men’s applications.
Clearly, pipeline improvements are not a fool-proof solution on their own. Inclusion must be built through a comprehensive approach to transformation, which includes addressing bias in data itself.
AI design processes should use datasets that reflect the populations likely to use or be affected by the technology. Collecting representative data involves measuring relevant characteristics, developing a nuanced understanding of the populations involved, and ensuring experts assess data classes and labels for new biases. Throughout this, methods for selecting and cleaning data should be documented to inform improvements; human oversight is critical to avoid mistakes and to ensure equitable decision making. Before release, models should be exposed to a wide range of real-life environments, contexts and users to ensure inclusivity.
The Call to Action’s third theme: ‘Engage your suppliers.’
Strengthening pipeline & practices and building solid data foundations are important for developing inclusive AI internally. For many organisations that purchase AI externally, however, these may not be feasible due to limited control over the biases within systems.
In 2020, 6/10 firms in Europe sourced their AI technologies externally through purchase of software and/or ready-to-use systems.[1] Engaging suppliers is a key opportunity to cascade inclusive AI throughout value chains and contribute to ecosystem transformation, while avoiding bias within purchased AI.
As the market of AI suppliers rapidly expands, the likelihood of biases within AI value chains also increases. More companies should start to share best practices and create AI Supplier Code of Conducts to maintain high standards of ethical and inclusive AI with suppliers.
Conclusion
At this critical moment, it’s time to embrace our responsibility and the urgent imperative to promote inclusive AI; to reap its full potential as a solution and to avoid exacerbating the crises we are working to solve. With united action, we can leverage inclusive AI as a powerful tool to accelerate transformation towards a more sustainable, equitable world for everyone.
The Women4AI Daring Circle is inspired and mobilised by its community of companies, organisations and experts who are engaging around this critical opportunity to drive inclusive digital transformation. Amidst the rapidly evolving world of AI, the group remains committed to accelerating uptake of inclusive AI and amplifying the Call to Action’s potential for impact. Moving forward, the group is excited to see signatories translate their commitments to action, and looks towards bringing more companies along this journey.
Read the full Call to Action and contact sophie.lambin@womens-forum.com if you would like to find out more and take part as a signatory.
[1] European Commission (2020), ‘European enterprise survey on the use of technologies based on the use of AI
The world is waking up to the wealth of opportunities to leverage Artificial Intelligence (AI) to address the converging crises we face as a global community; from climate change, economic inequality and war to societal polarisation and growing distrust.
This call to action invites organisations to commit to implement inclusive AI, whether that means embarking on first steps, accelerating existing initiatives or scaling best practices.
The use of artificial intelligence offers great business opportunities, but also triggers significant risks, such as unintentionally embedding biases in the decision-making process, as well as breaches of privacy, labour, and human rights.
So what are the strategies that leaders within organizations and businesses can take to leverage language for inclusion and equity in and through AI systems?
Finance is at the heart of the economy, and the opportunities that digitisation and AI provide will help to make finance much more accessible to all and simpler for everyone.
What role can AI have in bridging the disability divide to increase disability employment?
Right now, discussion about the need to reduce bias in technology is everywhere. The question now should not be only about how we can reduce bias, but how we can actively make AI inclusive.
As the world transitions to a more just and sustainable economy, technology is a key lever of change. Artificial Intelligence (AI) stands at the forefront of this transformation.
Many institutions make decisions based on artificial intelligence (AI) systems using machine learning (ML), whereby a series of algorithms takes and learns from massive amounts of data to find patterns and make predictions. Yet gender bias in these systems is pervasive.
A better representation of women and diverse mindsets in the development and deployment of AI would reduce the gender stereotypes and the negative consequences they lead to. If that is not the case, evidence suggests that by 2022, 85% of AI projects will deliver erroneous outcomes due to bias.