PIO model
P
Principles
7 ethical principles
I
Indicators
70 questions
O
Observables
Responses
PHASE 1
Presentation of the principle through various indicators
PHASE 2
Collection of all observable responses
PHASE 3
Evaluation based on the percentage of questions collected
PHASE 4
Visualisation of the result through the OEIAC badge
First, organisations will see a presentation of each principle and the selection of a set of question indicators strategically selected after a thorough review of the literature (Phase 1). Secondly, all affirmative and negative answers from the quick and full self-assessments will be collected as well as justifications to the answers for the latter, if applicable (Phase 2).
With the information obtained from the self-assessment, metrics will be obtained for each of the 7 principles assessed and with them observable profiles of each of the organisations that carry out the self-assessment (Phase 3). Once the corresponding metrics have been obtained, a badge or seal of quality will be created, which will be the graphic image of the PIO self-assessment model representing the adequacy of the data and the AI system according to the 7 ethical principles of the model: (1) transparency and explainability, (2) justice and equity, (3) safety and non-maleficence, (4) responsibility and accountability, (5) privacy, (6) autonomy, and (7) sustainability. A colour gradient is used in this process to quickly, simply and effectively represent the results obtained through the self-assessment model (Phase 4).
In this sense, before carrying out the self-assessment, it will be important to choose the person or persons responsible within the public or private organisations that use AI systems to carry out, from a competence point of view, a self-assessment with the maximum information available. To do so, binary answers (Yes/No) to 70 questions on each of the 7 principles described above will be collected.
A rapid self-assessment of position and a full self-assessment of position with recommendations may be carried out. The latter requires justification for yes or no answers, as this will allow the OEIAC team, together with external partners, to analyse the qualitative responses and provide feedback to the organisation that has carried it out, if desired. It should be noted in addition to obtaining specific metrics and a quality badge, organisations that have completed the full self-assessment will receive specific recommendations for improvement.
Rapid self-assessment: PHASE 1 + PHASE 2 + PHASE 3 + PHASE 4
Complete self-assessment: PHASE 1 + PHASE 2 + PHASE 3 + PHASE 4 + Feedback OEIAC
Finally, the badge or seal of quality is a graphic image of position based on the metrics obtained about the orientation and adequacy of the data and AI system to the different ethical principles according to the PIO model. In this sense, the different principles are placed inside the badge with a colour gradient which refers to the metrics obtained, with 0 being equivalent to no affirmative answers and 100 to all affirmative answers.