Driving Artificial Intelligence in Public Services through Accredited Upskilling

This information was sourced from "Driving Artificial Intelligence in Public Services through Accredited Upskilling" by Observatory of Public Sector Innovation (OECD), https://oecd-opsi.org/innovations/ai-certification-ireland/ Licensed under Creative Commons Attribution-ShareAlike 3.0 IGO (CC BY-SA 3.0 IGO).

Innovation Overview

Situation: Ireland intends to become a world-leader in AI under the National AI Strategy: AI, Here for Good, innovation stratgey "Making Innovation Real" and Civil Service Renewal 2030. This cannot be done without embedding AI across throughout the Civil and Public Service and its application into enhanced service delivery is trusted by the citizens of Ireland.

Task: To identify and advance the mechanism to develop new technological skills (through upskilling/reskilling) for Irish Civil and Public Servants, whilst also identifying a pipeline of potential, realisable projects that will benefit from strategic investment.

Action: In 2021, the Transformation Delivery Unit, DPER completed research to identify the appetite and need for AI upskilling/reskilling in Civil and Public Service.

To achieve the overarching task of advance potential AI projects, it was decided to make participation contingent on applicants submitting potential AI projects. Identify potential providers, secure budget, seek interventions that are accredited, appoint provider and secure value for money by aggregating demand through DPER.

A decision was made by DPER to cover participant cost (circa. €1,500 per participant). Public service organisations were required to provide senior level sponsorship to their participant’s time commitment and agree to their AI project.

A key action undertook by DPER was to communicate the Foundation Cert through established networks across the public service, allowing for the recruitment of participants. An evaluation model was developed to select participants on basis of application and project’s potential with expert in AI being implemented into reality.

Result: DPER, Transformation Delivery Unit has developed and secured a course, which is pitched at level 8 on the National Framework of Qualifications and is accredited by Technological University Dublin through a partner provider. Each course is delivered remotely with 60 hours of online classroom time and ongoing assessments

To date, 41 civil and public servants have been successfully selected, upskilled and 41 implementable projects (assessed by an expert in AI from the University College of Dublin) have been identified for follow-up by DPER Transformation Delivery Unit. A further iteration of the course is due to advertised and launch in Q4 2022, however based on the impact of the previous two iterations of the course delivery, each course has been oversubscribed demonstrating that the demand analysis was accurate.

A further iteration of the course is due to advertised and launch in Q4 2022, however based on the impact of the previous two iterations of the course delivery, each course has been oversubscribed demonstrating that the demand analysis was accurate.

Feedback from previous participants has been exceptionally positive and some participants have continued further studies in the field of AI (e.g. MSc. in AI).

What Makes Your Project Innovative?

This was the first accredited course of its kind geared towards Irish public servants. It was designed so non-technical persons can apply (although primary degree of some type required). Participants are introduced to the basics of planning and programming an AI solution by completion of the course. By coordinating the course centrally, additional oversight of AI projects across the Civil and Public Service can be maintained and enables cross-pollination and scaling of projects across the whole public service. This has been achieved to date by building a repository of projects completed, in planning and verified as suitable for AI deployment (for future follow-up). The course is building a directory of newly skilled AI-proficient public servants to embed this technology across our public service. It acts as a mechanism for students to find funding calls like the Public Service Innovation Fund or the European Digital Innovation Hubs to financially support their project development.

What is the current status of your innovation?

While the intention was never to educate people to the extent they could design, code, test and deploy the AI solutions to business problems themselves, the course we designed delivers on the goal of having Public Servants sufficiently educated in AI to see the business potential of the technology. Students can sit with private sector consultants and take a meaningful part in meetings where solutions are outlined. While developing AI project proposals, those who have competed the course are also empowered to adopt a Project Governance and Management role during system design, testing and deployment. This allows Public Servants to take real ownership and accept full responsibility for the quality and suitability of the AI solutions deployed. When a sufficient number of AI projects have been delivered, we plan to host webinars to showcase both successful and unsuccessful AI projects further strengthening the growth of AI across the Irish Public & Civil Service.

Collaborations & Partnerships

Project delivered: Solas - A ChatBot to handle course queries

Projects in planning: DSP - Assessing claims for fraudulent applications. DETE - Natural Language processing algorithms to extract bulk data.. Teagasc - Using AI to determine the best milking strategy for individual cows. Met Eireann - Measuring the amount of artificial light emitted from Ireland. HSE - AI powered chat-bot to deliver expert out-of-hours advice to assist parents of children with autism access services.

Users, Stakeholders & Beneficiaries

AI systems deployed in the provision of public services needs to be designed and operated in a manner that both ensures non-discriminatory delivery of services and protects the data of the citizen from inappropriate use or disclosure. Ensure that Government Officials ) have sufficient grounding in the essentials of AI, supporting collaboration with private sector suppliers in the development of AI trustworthy based solutions that provide easier access to public services digitally .

Results, Outcomes & Impacts

While this initiative is relatively recent, many projects are now being actively explored, whilst one project has successfully delivered and exceeded expectations. To monitor the effectiveness of the course to grow both the awareness and capability of public service organisations in the use of AI in the delivery of public services, we survey the participants after the course and where allowed by the students, publish the results. This has further heightened awareness and enthusiasm within public service organisations to send their people on the course. A course is currently in process with 20 students and the next course is already over-subscribed. By having a closed course for Civil and Public Servants, ideas can be shared with each other triggering enhanced service delivery projects in their respective organisations and encouraging cross-collaboration with projects. This project has successfully created a register of skilled professionals in AI and a repository of AI projects.

Challenges and Failures

The Foundation Cert AI has experienced an oversubscription of course applications. This has been overcome by prioritizing unsuccessful applicants (with appropriate projects) for future iterations of the course.

The qualifying educational requirements has resulted in potential applicants not being eligible for course. The project team has overcome via upfront communications stating this is a requirement of TU Dublin(awarding body).

There has been difficulties in securing appropriate experts in this field (e.g. project peer reviewer), which has been generally resolved through persistence in rescheduling dates in the course calendar.

Securing the funding to support multiple iterations of the course has been challenging. This has mostly been overcome through robust business cases to senior management. Funding has been secured to run a further two courses in 2022/2023.

Conditions for Success

Prior to beginning the project, key stakeholders were engaged from the OGCIO, DETE (responsible for National AI Strategy), European Commission (AI Accelerator programme) and the OECD's Observatory on Public Sector Innovation to ensure alignment with national and supranational policy and best practice. As part of the structured market research and engagement exercise, key educational providers were researched to understand different curricula, models of delivery and assessment, eligibility and restrictions, focus on ethics, costs, accessibility for non-technical persons etc. Interest in the Foundation Cert in AI was gauged from the wider Civil and Public Services, through engagement with internal networks Innovation Networks (2000+ members). The project team identified one of Ireland's leading experts (Director of CeADAR, Ireland's centre for applied R&D in AI, Machine Learning and Data Analytics based in UCD) to support in the assessment of applications.

Replication

The project is entirely aimed at introducing Civil and Public Servants to emerging technology and the application of AI in the delivery of public services. As part of the selection process, the team seek to ensure balance of participants to ensure spread across the Iris Public Service. A key feature of the project is to encourage the initiation of projects within applicants' organisations and this allows DPER to identify projects that have scaling potential and replicability potential. This will enable value for money by preventing a fragmented approach to project creation ('mushrooming'). The course model has the functionality to be replicated by other public and civil service bodies (especially larger ones) that may seek to run internal AI foundation courses. All Information on the curriculum, procurement process and lessons learnt can be easily shared to any organisation, which actively seeks to develop Internal skillsets.

Lessons Learned

This course contains a considerable amount of active participation by the student. There are data-analytics and programming elements where students are expected to complete some group and solitary projects. As such it is not possible to just attend and get the qualification on completion. There are actual course deliverables that are formally marked by an accredited academic institution. While this places extra demands on the student, the feedback on completion of the course strongly indicates that the 60 hours of time invested by both the student and their host organisation, delivers what we wanted from the course. The above gives clear guidance as to how training of this nature needs to be structured. The experiential element of the course is critical not only in terms of student enjoyment, but also in terms of the student have a strong sense for the difficulties, delays and challenges of deploying these technologies in a real-world scenario.

Anything Else?

The title of the Irish Government’s AI strategy document “AI: Here for good”, hints at two things simultaneously. AI is here to stay and Governments need to use its potential to deliver public services more economically. AI needs to be implemented ethically so that citizens can derive the full benefit of more accessible and new services that AI can enable. The Irish Civil Service in the strategy document Civil Service Renewal 2030 Strategy - 'Building on our Strengths', states that were appropriate 90% of public services will be delivered digitally. This will require not only AI educated Civil and Public Service personnel, but also public servants who are confident in its ability to use this technology. This AI competence will not happen by itself. Courses such as the one detailed in this document are key to developing our own people so their current domain expertise can be utilised in the development and deployment of AI enabled public services.