Approaches and use of ai in the public sector by the european member states analysis of national ai strategies(1)
1. 18.07.2021
Colin van Noordt
PhD Researcher, TalTech
8th International Summer School on Digital Government
APPROACHES AND USE OF AI IN THE
PUBLIC SECTOR BY THE EUROPEAN
MEMBER STATES
ANALYSIS OF NATIONAL AI STRATEGIES
2. TALLINN UNIVERSITY OF TECHNOLOGY
WHO AM I?
▪ Academic background:
▪ MSc Public Sector Innovation & eGovernance (KU Leuven, Münster
University & TalTech)
▪ BSc Pubic Administration, Leiden University
▪ PhD research topic: the Public Value of AI in Public Services
▪ External Expert to the European Commission’s AI Watch research team
on AI in the public sector
▪ Research activities:
▪ Understanding the concept of AI in government and refining the
concept based on empirical use cases
▪ Requirements needed for using AI technologies in government
organizations
▪ Policy initiatives and strategies for facilitating AI in government
▪ Impact and consequences of AI deployed in public services
▪ Governance of AI technologies and ensuring responsible use
3. TALLINN UNIVERSITY OF TECHNOLOGY
TOWARDS AI IN GOVERNMENT
● Governments are often seen as a vital player in assisting the
development and uptake of AI in society
● Triple (or even more) roles governments are expected to take
As Facilitator
• Opening public
sector data,
establishing
data
collaboratives,
stimulating
GovTech
As Regulator
• Ensuring
framework
conditions and
limiting
negative
consequences
of AI
As User
• Deployment of
AI by the
government to
tackle social
challenges
4. TALLINN UNIVERSITY OF TECHNOLOGY
IMPLEMENTATION OF AI IN
GOVERNMENT
▪ There is great interest from policy makers as well
as academia on the use of AI in public
administration
▪ A recent study (Zuiderwijk et al. 2021) highlight
the various implications (positive and negative)
of AI in government as highlighted by literature
▪ Despite this (massive) interest, public
administrations are not using AI to their full
potential due to the various adoption and
implementation barriers (see e.g. Sun &
Medaglia, 2019 Wirtz, Weyer & Geyer, 2019)
5. TALLINN UNIVERSITY OF TECHNOLOGY
RESEARCHING NATIONAL AI
STRATEGIES
▪ Research is increasing on analysing AI
strategies
• Policy areas (Fatima et al., 2020;
Valle-Cruz et al., 2019)
• Textual patterns (Dexe & Franke,
2020)
• Public values (Robinson, 2020; Toll et
al., 2019; Viscusi et al., 2020)
▪ Development of National AI Strategies
• Part of the Coordinated Action Plan
• To coordinate and share
implementation measures on AI
6. TALLINN UNIVERSITY OF TECHNOLOGY
ANALYSING NATIONAL AI STRATEGIES
Collection of
published* AI
National
Strategies
(n=21)
Excluded strategies in
draft of concepts phase
Excluded other policy
documents, reports or
expert group strategies
Identification of
passages
AI for use by
public
administration
Requires clear
reference to the public
administration use
Describes actions,
initiatives or
suggestions to
facilitate, stimulate or
reinforce the
development and
uptake of AI in public
administrations
Summarization
and analysis
Identification of
frequently mentioned
policy
Comparison and
evaluation of initiatives
*As of April 2021, there are 21 countries which have published their national AI strategies: Bulgaria, Cyprus, Czec
Republic, Denmark, Estonia, Finland, France, Germany, Hungary, Latvia, Lithuania, Luxembourg, Malta, The
Netherlands, Norway, Poland, Portugal, Slovakia, Spain, Sweden, and the United Kingdom.
7. TALLINN UNIVERSITY OF TECHNOLOGY
POLICY INSTRUMENTS TO BOOST AI ADOPTION
Stimulating awareness
and knowledge sharing
Strengthening data
management for AI
Improving data quality
Improving data access
Access to private sector data
Aims to improve the data
landscape of the public sector
to facilitate learning of AI
The Lithuanian strategy
highlights introduction of data
management plans
Awareness campaigns
Regular meetings
Policy events
Helps civil servants
understand AI and detect
use cases in their work
Hungary will improve
awareness of AI for public
officials
8. TALLINN UNIVERSITY OF TECHNOLOGY
POLICY INSTRUMENTS TO BOOST AI ADOPTION
Building internal capacity Learning by doing
Generalist training
Specialist training
New positions
Focus on enhancing the
skillset of staff to
develop and use AI in
their work
In Slovakia, Digital
Innovation Laboratories
are to be created for
innovating with AI
AI pilot projects
Regulatory sandboxes for
AI
Countries are
experimenting with AI to
learn and share
knowledge.
In Norway, public
administrations are
allowed to test new ways
of working up to 4 year
9. TALLINN UNIVERSITY OF TECHNOLOGY
POLICY INSTRUMENTS TO BOOST AI ADOPTION
Ethical and legal
framework
Ethical framework
Reform of data sharing
laws
Specific AI law
Action aimed at providing
ethical and legal guidance
for AI use
In Spain, an ethical guide
will be developed to assist
civil servants
Funding for AI in the public sector
Stimulating GovTech & Incubators
Revising public procurement
Stimulate AI through funding
and procurement
Cyprus suggests
implementing tax relief and
financing for public
organizations testing AI
Ethical and legal
framework Funding and procurement
10. TALLINN UNIVERSITY OF TECHNOLOGY
OVERVIEW OF ACTIONS AND FREQUENCY (N=21)
Stimulating
awareness
and
knowledge
sharing
Awareness
campaigns
(12)
Organising and
hosting
meetings
(7)
Participating in
and promoting
events
(10)
Strengthening
data
management
for AI
Enhancing
data quality
(19)
Improving
accessibility to
public data
(20)
Access to
private sector
data
(7)
Building
internal
capacity
Generalist AI
training
(12)
Specialist AI
Training
(8)
New positions,
institutions or
departments
(11)
Learning by
doing
Pilots,
experiments
and flagship
projects
(15)
Regulatory
sandboxes for
AI
(11)
Ethical and
legal
framework
Development
of an ethical
framework (14)
Reform of data
sharing laws
(11)
Possibility of a
specific AI law
(2)
Funding and
procurement
Funding for AI
projects in the
public sector
(10)
Stimulating
GovTech &
incubators
(12)
Revising
procurement
processes (9)
Other
Reusable AI
solutions and
platforms
Improving IT
infrastructure
Changing
project work
practices
11. TALLINN UNIVERSITY OF TECHNOLOGY
CURRENT KEY FINDINGS
▪ Relative lack of clarity regarding the depth of scope of initiatives
▪ Strategies often blur concrete policy actions, ongoing initiatives and ‘wishes’/ ‘nice-to-have’s’
▪ Strategies focus strongly on overcoming data-related barriers to AI
▪ Organisational factors and resources needed for AI may be overlooked
▪ Close the ‘gap’ between the private and the public sector
▪ Strategies describe many more actions to facilitate at the private sector than public sector
▪ Improving and boosting public procurement for AI is promising
▪ However, a successful AI procurement still requires internal capacity and skills. Focus on supply and demand side of
the procurement process. Overcoming risk-averse culture is crucial to in the end use the opportunities provided.
▪ More funding for AI in the public sector is needed, and made more specific
▪ Not just for research and development of AI, but for piloting and introducing organisational changes
▪ Relatively unclear what causes these differences in the national AI strategies
▪ Potentially historical eGovernment development, administrative cultures or other.