Three Ways to Embed DEI Into Your Company’s AI Strategy

Originally published  on HBR.org / January 08, 2024 / Reprint H07YS6
By Andy Baldwin

What/focus

There is wide  consensus that Artificial intelligence (AI), and particularly generative AI specifically, is a game changer that will drive innovation, productivity, and revenue. However, AI cannot be a force for good if it serves to exacerbate existing inequalities in the workplace and society more broadly. Efforts are being made globally to establish guidelines, regulations, and ethical frameworks for responsible AI to address bias, and build transparency, accountability, and fairness into AI systems. Businesses will increasingly be expected to deploy automation in
responsible ways to maximise its potential, but not at the expense of widening the gap between those who have always had access to opportunities and those traditionally disadvantaged, or who may even be displaced by AI.

This article discusses balancing DEI (diversity, equity, and inclusion) with automation in the deployment of AI systems as a business imperative, and outlines three ways of doing this.

 How (details/methods)

The first strategy is to embed DEI into the design of AI systems to counter the risk of algorithmic bias, for example in job screening, mortgage decisions, and health care. This requires careful design, testing and guardrails. An important step is involving a wide range of stakeholders from different demographic groups and backgrounds to design in diversity and wider access. There should be diversity in the testing team as well. Wider input and access will in turn lead to better data to train the system for the real world.

Second, DEI should be incorporated into and prioritised in upskilling and training programmes. Equipping  underrepresented groups with critical AI skills will lead to more responsible AI systems. It will also enable individuals to transition into new and better paid roles. AI and machine learning specialists are currently the fastest growing job category relative to their size. People at all levels of an organisation need to understand how to deploy, use, and manage AI tools, and this includes being aware of algorithmic bias.

Finally, strategy 3 involves using AI to directly boost DEI by identifying patterns that negatively impact certain groups, for example pay gaps within an organisation. Identifying such blind spots can ultimately support more equitable performance evaluation, promotion processes, resource allocation and project assignments. AI can also be applied to tailor inclusion initiatives based on employee engagement, satisfaction and feedback. Strategy 3 also encompasses improving workplace accessibility for people with disabilities, through speech recognition and image recognition tools for example.

So what

Broadening input into AI design to create wider access and applicability, prioritising  more equitable access to upskilling, and applying AI directly to boost DEI will lead to people-centred AI transformation. This will achieve a balance between automation and DEI as AI reshapes the nature of work.