Principles to Guide Engagement Around AI For Scotland

We are very pleased to share the report titled ‘Design principles and mechanisms for community engagement for Scotland’s AI strategy’ which details the process and insights from co-design work carried out by Democratic Society (DemSoc) through Winter 2022/23.

The Background

At the Scottish AI Alliance, we hold a vision to make Scotland a leader in the development and use of artificial intelligence which is trustworthy, ethical and inclusive. A big part of that is to work with the people of Scotland to boost public awareness and understanding of AI and how it impacts their lives, and to actively seek new voices from groups who are under-represented and marginalised in Scottish society.

To deliver our public and community engagement in a way which is trustworthy, ethical and inclusive we wanted to be guided by strong principles which meet the needs of various publics and communities. To ensure that these principles were representative of and responsive to the people of Scotland, we commissioned DemSoc to deliver a series of workshops with participants drawn from communities across the country.

Our approach has been to consider engagement and participation holistically, believing that no one single mechanism for engagement or participation will address the needs of the people of Scotland around AI. As well as discussing the principles that should guide our engagement, our workshops explored how to decide the best engagement mechanisms for different communities to ensure that our logistical approach will be bespoke and appropriate with everything we do.

The Workshops

We kicked off our workshops in Leith, with a day-long in-person session at Out of the Blue Drill Hall in November 2022. We followed this up with a day online with participants from across Scotland in December 2022 and topped this off with another in-person session in Inverness in January 2023. In total we worked with 35 members of the public across the three workshops.

At each workshop we discussed the following questions:

What is AI?

Almost all participants had an understanding of what AI is, and what it is not. We established a baseline of terminology for later discussions, and gained insight on public perception of AI.

What are your AI hopes and fears?

Hopes around AI included AI technologies helping improve healthcare services and boosting workplace efficiency. Fears included the prospect of job losses and the impact of algorithmic bias and discrimination.

What does ‘engagement’ mean?

Participants shared that they understood engagement as being a process of listening, connection and communication which is collaborative and accessible.

What does ‘public’ mean?

Discussions around ‘the public’ focussed on ensuring that the audience of engagement is truly inclusive of age, location, access to technology, level of education, language and background.

Principles of Engagement

Once we had established this baseline of knowledge on AI and public engagement, we looked at different scenarios involving AI technologies as a group, using a fictional narrative drawn from real-life examples of use cases. When considering each scenario, our participants were asked to consider the scenario and discuss who should be engaged in that illustrative example, and how they should be engaged. The scenarios ranged from low-stakes, low-impact examples to high-stakes, high-impact, and participants were asked how the difference in stakes and impact changed who and how people should be engaged.

From this we guided the discussion towards what principles would help us achieve best practice in everything above.

Common themes emerged, and the following principles were defined by this co-design process:

The Principles

I.                 Endeavour to Overcome Assumptions and Biases

Actively and routinely attempt to identify and overcome underlying assumptions and biases on the topics and techniques of engagement and AI.

II.               Meet and Enable Participating Publics Where and How They Are

Creatively tailor engagement experiences which are fit-for-purpose to help participants learn, inquire, critique, ideate, and make decisions on topics and techniques of AI.

III.              Foster Inclusive Processes for Diverse Publics

Ensure respectful inclusion and accessibility for the wide variety of diversely affected experiences and perspectives on the topics and techniques of AI.

IV.             Create Transparent and Traceable Engagement Accountability

Establish transparent and traceable practices including feedback loops for ongoing accountability between participants and decision-makers on the topics and techniques of AI.

The Next Steps

While we build our engagement process in line with this report we will be posting further blogs on how we will deliver against our co-designed principles, and the insights we gain from doing so.

In the meantime, you can read the report itself by clicking the button below.