BUILDING INCLUSIVE AI: HOW CAN ORGANISATIONS TAKE ACTION?
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, holding immense potential for innovative, long-term growth. Valued at 93.5 billion USD in 2021, the global AI market is projected to expand by almost 40% by 2030.
There is no denying that the power of AI is growing rapidly and impressively – but how can we ensure that its growth is inclusive?
As both the private and public sectors explore the breadth of new opportunities created by AI, there is growing recognition that new technologies can (and must) be leveraged to shape a more inclusive future and mitigate emerging social risks. The Women4AI Daring Circle convenes key actors from business, policy, science and civil society to collaboratively harness the power of AI to drive inclusion, both in its development and its application.
In June, Daring Circle members engaged in an intensely dynamic and constructive workshop discussion on best practices for Inclusive AI. This article outlines some of the many insights and actionable steps that emerged from the discussion, highlighting the vast opportunity for business to promote and advance inclusive AI.
Best practice & opportunities for organisations to take action:
- Make inclusion a priority: Due to the relatively recent rise of AI, it is important to embed inclusive values as a foundational exercise, rather than an add-on exercise. Innovating a company code of conduct for ethical AI, informed by national policies and current research, can ensure that internal codes work within the broader ecosystem. Shaping strategy around this code ensures that objectives are aligned and employees are on board to facilitate innovation which promotes ethical AI.
- Build internal capacity: Scaling and opening up training, even at the general level, can help bring in more perspectives and increase awareness of bias. Including academic research is crucial; technical topics, such as algorithmic auditing and certification, can complement broader training around social and cultural biases.
- Tackle bias strategically: Technology is not always neutral – when Machine Learning enters our institutions and structures it can learn systemic biases. Thus, Inclusive AI programmes will inevitably face problems with bias. As there is no single way to measure, define, and identify its source, informed strategic thinking is crucial to develop guidelines around bias. Focused internal conversations and effective feedback loops are key ways to ensure bias mitigation strategies are informed by and embedded within the organisation. It is important to go beyond anti-bias training to understand implicit bias.
Promoting Inclusive AI in the broader ecosystem:
- Share resources: There are distinct information gaps between technology users and creators – addressing this by amplifying knowledge and sharing data can improve trust and lead to greater uptake of inclusive technology. To support the development of an inclusive AI across the ecosystem, share available resources; for example, the Women4AI Action Toolkit can help others navigate this important, yet unknown, space.
- Join the conversation: Amplify the importance of the issue and share knowledge by organising and participating in conversations, conferences, working groups and academic networks. Build internal and external awareness by communicating commitments to customers, business partners and collaborators. Engage sales staff at technology companies to raise ethics as a business imperative especially while talking to customers.
- Amplify marginalised voices: Including diverse perspectives in the development and testing of technology is crucial for both building effective, inclusive AI and mitigating its potential to exacerbate existing inequalities. Amplify voices of vulnerable communities, while ensuring that compensation and credit is shared fairly in a manner that empowers communities and individual contributors.
- Improve access to data: Open access to data decentralises power and knowledge; allowing researchers access to data for research purposes can increase support and understanding of user missions.
With the AI space growing and changing rapidly, having a clear and strong vision with inclusion at the centre is crucial. Partners discussed their ideas of what successful, long term business engagement around Inclusive AI looks like. A compelling vision emerged, rooted in responsibility to advance learning and promote responsible AI approaches; regulation and governance that encourages inclusive innovation; and urgency to take action that ensures the development of AI consistently builds fundamentally inclusive technology.
There is an arms race between law and AI. The glacial development of law around AI is failing to meet the pace of emerging technologies’ upgrades and updates. As policy advisors reiterate the known prevalence of ethical AI, business can help build trust and respect into human to machine relations.
As AI matures, its growth continues to present limitless applications as an efficient tool which can facilitate cutting edge solutions to business problems. The Women4AI Daring Circle convenes around a shared desire to identify solutions and prevent human bias from replicating in AI through open, cross-sectoral discussion. The Daring Circle is releasing a 2022 Call to Action which will urge organisations of all sizes, from all sectors, to make concrete, process-based commitments that are reflective of the changing AI environment. The Call to Action, which focuses on inclusion beyond representation, will launch at the Women’s Forum November Global Meeting in Paris – contact email@example.com if you would like to find out more and take part as a signatory.