The AI ethics dashboard for pilots

Eight questions, three ratings, one form per pilot. The template that turns AI principles into something a region can check.

What does the PoliRural+ AI ethics dashboard ask a pilot to do?

For each of the eight principles the pilot answers a guiding question about its current practice (for example how it handles bias, limitations or hallucination) and rates itself as fully integrated, needs improvement or not implemented. One form per pilot; a companion form collects one local story of up to 250 words per principle.

Official title: PoliRuralPlus AI Ethics & Participatory AI Maturity Tracker (WP2)
Eight guiding questions, three ratings: fully integrated, needs improvement, not implemented
A story form asks for one local story per principle, up to 250 words
The filled rows in D2.3 are labelled fictionalised illustrations, not results

Open in the interactive book · missions, mini-games and memory cards there

The AI ethics dashboard: each pilot rates itself, principle by principle

main tellingdepth: standard3 min markdown

D2.3 proposes a self-assessment for every pilot region, the PoliRuralPlus AI Ethics & Participatory AI Maturity Tracker. Its stated objective is to standardise how each pilot reports its alignment with the Participatory AI Principles.

What the form asks

One form per pilot. For each principle the pilot describes its current practice in answer to one guiding question, then picks a rating: Fully Integrated, Needs Improvement or Not Implemented.

Principle	Guiding question (D2.3, Table 70)
Transparency & Explainability	How do you explain AI decisions to users?
Stakeholder Co-Creation	How were stakeholders engaged in tool or policy design?
Multi-Actor Involvement	Who is involved across the quadruple helix?
Iterative Feedback	How is stakeholder or user input fed back into tools or policies?
Ethical & Responsible Use	How do you address issues like bias, limitations or hallucination?
Capacity Building	What training or support did you offer to stakeholders?
Inclusive Policy Design	How was AI used to shape foresight or RAPs inclusively?
Gender Equality & Social Justice	Gender and other diversity indicators for social justice (a field description, not a question)

The overview table then gives each pilot an overall status. The template uses In Progress, Mature and Developing.

A companion form, “AI in Action for My Region”, collects evidence: the pilot region, the principle featured, a title, a local story of at most 250 words, an optional stakeholder quote and an optional image. The aim is one story per principle per pilot.

Illustrations versus results

D2.3 fills the template with sample rows: a Monaghan row marked “needs flagging bias issues”, a Slovak row with “feedback not looped yet”. D2.3 itself labels these fictionalised summaries that show how pilots might submit content. They are not assessments.

The real material is in D2.3 §2.8. Pilots submitted six “AI in Action” stories through an online form, from Päijät-Häme (Finland), Vidzeme (Latvia), Slovakia (two stories), the Czech-Bavarian border and Monaghan (Ireland). D2.3 reports that the stories cover seven of the principles and three services: the Advisor, MAAT and custom GPT-based indicator tools.

Why a form helps

D2.3 also proposes linking the principles to project KPIs: capacity-building KPIs to principle 6, policy co-creation outcomes to principles 2 and 7, and stakeholder feedback mechanisms to principle 4. A rating means something only with the practice behind it, which is why every row asks “how”.

To start in your own region, fill the rows where you have a story first, then take the weakest remaining row as your next task.

Filling the ethics tracker for County Monaghan

another tellingexample world: irelanddepth: standard2 min markdown

The Monaghan pilot submitted one “AI in Action for My Region” story, filed under the Stakeholder Co-Creation principle. Here it is put through the ethics tracker row by row. The ratings are left blank on purpose: only the pilot can give them.

The story in brief (D2.3 §2.8, Story 6)

The team began with a desk study of the circular economy in Monaghan, using the Advisor to map local challenges, relevant policies and stakeholders. It shared a roadmap at a co-creation workshop with over 20 regional stakeholders from business, education and local government. The Advisor then helped analyse the feedback, identify funding opportunities and develop three EU project proposals, and it helped define the interventions and actions of Monaghan’s Regional Action Plan for the circular economy. In parallel, the Outreach project GREEN scoped circular business opportunities and mobilised 23 key stakeholders, 21 of them SMEs (D6.3).

Row by row

Principle	Evidence on the public record	Still for the pilot to answer
Stakeholder Co-Creation	Workshop with 20+ stakeholders; RAP actions built on their input	How the Advisor’s drafts were changed after the workshop
Multi-Actor Involvement	Business, education, local government; GREEN reached 21 SMEs	Where civil society sat at the table
Transparency & Explainability	Not described in the story	How participants saw which sources the Advisor used
Iterative Feedback	Advisor used to analyse workshop feedback	Whether stakeholders saw and corrected the result
Ethical & Responsible Use	Not described	How errors and bias in AI drafts were caught
Capacity Building	Not described	Whether stakeholders were trained to use or question the tool
Inclusive Policy Design	Community priorities captured in the RAP	Which groups were missing
Gender Equality & Social Justice	No disaggregated figures in the story	Participation by gender and other indicators

What the exercise shows

One honest story fills two or three rows well and leaves the rest open. That is the point of the template: the empty cells are the pilot’s to-do list. D2.3’s own sample Monaghan row, a fictionalised illustration, marks ethical use as “needs flagging bias issues”. The real story gives no evidence either way, so the row stays open until the pilot describes its checks.

Builds on: Participatory AI Principles