6 steps to better conversations that can reimagine AI regulation – World Economic Forum

AI and algorithms are everywhere, underpinning many parts of our daily lives, improving systems and increasing productivity. They also carry risk, however, potentially introducing new biases or worsening old ones, blocking candidates from jobs and even misidentifying crime suspects.

As AI is ubiquitous, governments and businesses leverage social licence to explore new uses. Through social licence, communities agree that governments, agencies, or companies are considered trustworthy enough to use AI in ways that may have risks.

Having this trust means people believe that if something does go wrong, it will be quickly identified and fixed before there is harm caused. Such trust can be built with conversations and robust engagement between the public, technologists and policy makers.

Successful conversations on AI or other emerging technologies are multi-stage processes, involving different players at different intervals to play important roles. As part of the World Economic Forums Reimagining Regulation for the Age of AI project, the project team and community developed a range of best practices and steps for how to design a successful conversation on AI. For better, more robust conversations, these tips can help.

The COVID-19 pandemic and recent social and political unrest have created a profound sense of urgency for companies to actively work to tackle racial injustice and inequality. In response, the Forum's Platform for Shaping the Future of the New Economy and Society has established a high-level community of Chief Diversity and Inclusion Officers. The community will develop a vision, strategies and tools to proactively embed equity into the post-pandemic recovery and shape long-term inclusive change in our economies and societies.

As businesses emerge from the COVID-19 crisis, they have a unique opportunity to ensure that equity, inclusion and justice define the "new normal" and tackle exclusion, bias and discrimination related to race, gender, ability, sexual orientation and all other forms of human diversity. It is increasingly clear that new workplace technologies and practices can be leveraged to significantly improve diversity, equity and inclusion outcomes.

The World Economic Forum has developed a Diversity, Equity and Inclusion Toolkit, to outline the practical opportunities that this new technology represents for diversity, equity and inclusion efforts, while describing the challenges that come with it.

The toolkit explores how technology can help reduce bias from recruitment processes, diversify talent pools and benchmark diversity and inclusion across organisations. The toolkit also cites research that suggests well-managed diverse teams significantly outperform homogenous ones over time, across profitability, innovation, decision-making and employee engagement.

The Diversity, Equity, and Inclusion Toolkit is available here.

Whats needed for powerful engagements

Strong engagement forges trust and will power valuable conversations to help different parties fully understand each others needs and views. Good engagement takes several factors into account, including:

How to run a successful engagement

Planning can create the right spaces for strong engagements. To make the most out of these opportunities, the following steps should be taken:

1. Define. To start, the design team must identify core elements such as principles, legal context and societal norms underpinning the current use of AI. These current rules and conventions outline which boundaries, safeguards and protections are already in place. Communicating these elements to participants helps focus discussions by letting the public know which concerns are already covered by regulation and how participants can contribute to the creation of new safeguards.

2. Discover. In this stage, the design team researches the issue and the current context, identifying how the engagement could fuel progress. This initial stage could take place with a small-scale workshop involving a few key stakeholders. This stage allows the designers to crystallise the issue at hand, the aspects where consultation is needed, and any goals or outcomes. Participants will trust the process more if they can see the pathway ahead, know they can contribute to the desired end goal, and understand they are helping to design some of the steps to reach the end goal.

3. Decide. Here, the design team identifies who should be included in the larger engagement. This might include stakeholders as well as members of the community with particular concerns about, or knowledge about, bias and algorithms. Hard-to-reach audiences need to be identified and decisions made on how best to reach these participants. Designers will also decide what level of influence the participants will have and the extent to which views will be listened to, respected, and used in the next stages. Trust will be lost if people participate and then find their views have been ignored.

4. Design. In this stage, the team takes what it has learned about the issue it is covering, and the people involved to decide the best channels for engagement and dialogue. The shape of the engagement will depend on factors such as the intended participant list, the level of involvement being sought by the participants, and the nature of the feedback wanted. So, for example, a small-scale engagement with a specific group seeking views on how personal data is to be used, with the aim being to feed into a piece of legislation, will require different materials, meetings and level of discussion than will a general nationwide conversation on how human rights sit within digital technologies, with the aim of raising awareness about digital rights.

5. Analyse. Once the engagement has occurred and information gathered, the results need to be analysed and synthesised by the design team. The design team needs to look at the inputs and adjust for bias, vested interests, monopoly of views and strong voices. The findings will need to be presented to key stakeholders and other key audiences, including the participants. These findings will include recommendations on how the engagement research will be used (that is, it will inform or change the decision?) and the next steps.

6. Review. Finally, the whole process should be reviewed and evaluated, either by the design team or by an independent group. This will help identify what went well and what didnt, giving valuable insights for future engagements. The review can be done in many ways (For instance, interviews or surveys with the participants and design team is a useful way to review the process from both sides). A key part of this process is providing honest feedback to participants on what happened with their input and how it has contributed to the next steps.

"Through trust, its possible to yield the valuable insights and material that make the ultimate decisions stronger."

Open and honest engagement requires courage on both sides. Conversations on new technologies that have the potential to disrupt communities and economies are difficult. A carefully planned engagement helps signal transparency and humility to participants and the fact that one party has all the answers.

Engagements will inevitably result in differing views or roadblocks. The steps above cant prevent disagreements, but if the designers are open and honest from the start, trust is built early and its possible to work through any disagreements. Through trust, its possible to yield the valuable insights and material that make the ultimate decisions stronger.

An involved and active community that trusts enough to engage is vital to a functioning democracy. Gaining and maintaining this trust requires an ongoing commitment but one that is worthwhile as community support and input makes for better, stronger policy.