Ethical Issues on eGovernment 3.0: Big Data and AI

Ethical Issues in eGovernment 3.0: Big
Data and AI
Alexander Ronzhyn
University of Koblenz-Landau,
Nationales E-Government Kompetenzzentrum
ronzhyn@uni-koblenz.de

What is Ethics?
 Ethical behavior can be defined as “doing the right
thing, showing concern for people and treating people
right, being open and communicative, and
demonstrating morality in one’s personal life”
gov30.eu 2
(Treviño et al., 2000, pp.131-132)

Ethics in e-Government
Ethics in
Digital
Government
ICT Ethics
Ethics in
Government
gov30.eu 3

Causes of algorithmic bias
 The Design of the Algorithm
 The Code written to implement it
 The Data used to train it
gov30.eu 4
Social prejudices embedded
mathematically into algorithms
with widespread influence

Algorithmic bias in the wild
gov30.eu 5

Algorithmic bias in policy
 Predictive sentencing
 COMPAS algorithm used in the US to predict rate of
recidivism
 Scores were wrong 40% of the time
 Black defendants were often predicted to be at a higher risk
of recidivism than they actually were (twice as likely to be
misclassified as higher risk).
 White defendants were often predicted to be less risky than
they were (mistakenly labeled low risk almost twice as often
as black re-offenders).
 Bail amounts calculation
 Compared to white men, Latino men were asked to pay
19% higher bail. Black men were asked to pay 35% higher
bail.
gov30.eu 6

Methodology
Identify main
ethical concerns
in digital
government
Review literature
on ethical issues
in disruptive
technologies
Map the
identified issues
to the concerns
gov30.eu 7
22 articles
645 articles
74 articles:
• 27 AI
• 47 Big Data
Literature review:
Kitchenham and Charters
(2007)

Ethical considerations in e-government
gov30.eu 8
Inclusivity
Privacy
Data use
Quality/ Accuracy of information
Transparency
Accountability
Information ownership
Trust
Alignment of values
Cost

Inclusivity
 A concern about the inability of some groups of citizens
to make use of the digital government services. It is
usually discussed in the context of the digital divide
either within a society or between countries. Most
common factors causing digital divide are disparity in
access to technology, wealth, education or age-related
differences.
gov30.eu 9
(Mordini et al., 2009)

Privacy
 A concern about the unauthorized or inappropriate use
of the individual information by the Government or
other actors. Privacy is the most discussed ICT-related
ethical issue, especially after the advent of social media
and large-scale personal data collection.
gov30.eu 10

Data use
 A concern about the inappropriate use of collected data.
This includes for example the aggregation of data from
different sources to infer new information or to de-
anonymize individual citizens.
gov30.eu 11
(Mason, 1986)

Quality and accuracy of information
 A concern relating to the imperfect digitalisation of
certain government data during the transition to the
digital services. Data errors or incomplete information in
the databases may result in additional costs for a citizen.
gov30.eu 12
(Anderson, 2004)

Transparency
 A concern that certain processes in digital government
may become black boxes, impossible to understand by
individual citizens. Lack of transparency may lead to the
inequality of treatment, when certain decisions are made
using invisible decision processes based on data only
available to the system.
gov30.eu 13
(Henman, 2005)

Accountability
 Accountability is related to transparency and concerns
the responsibility of government toward an individual
citizen in case of problems with or misuse of the digital
government system. Accountability is necessary to
improve citizen trust in e-government.
gov30.eu 14
(Welch, Hinnant, & Moon, 2005)

Information ownership
 A concern about the possibility of the digital
government system’s user to change or restrict access to
one’s own information. It also concerns the re-use of
certain information from the e-government systems by
the third parties.
gov30.eu 15

Trust
 A consideration of the effect that the automatization
(and associated de-humanisation) of the government
services may have on an individual citizen. It also
encompasses the issues of government control and
surveillance.
gov30.eu 16

Alignment of values
 This concern refers to the mismatch between the values
of the government and the citizens. Sometimes
motivation of the government to introduce digital
services (e.g. cutting costs, improving efficiency) may
not be aligned with the interests of the citizens, who
value accountability and inclusivity of the services.
gov30.eu 17
(Berger, 2016; Fairweather & Rogerson, 2006)

Cost
 The cost consideration refers not only to the financial
cost of implementing and running the digital
government services but also trade-offs for the citizens,
associated with the implementation of e-Government
services: ensuring inclusive access to government
services may increase the workload for the civil servants
and thus the cost of public services.
gov30.eu 18
(Berger, 2016)

Results: Ethical concerns about AI in e-government
gov30.eu 19
Major issues
Accountability
Value alignment
Privacy
Inclusivity
Cost
Minor issues
Transparency
Trust

Results: Ethical concerns about Big Data in e-
government
gov30.eu 20
Major issues
Privacy
Data ownership
Data accuracy
Data use
Alignment of values
Minor issues
Inclusivity
Transparency and
accountability
Trust
Cost

Ethics in
Government
3.0
Ethical considerations in e-
government
Inclusivity Inequality between those who control AI and other people
Privacy
Surveillance and profiling.
De-anonymization of data through cross-referencing.
Data use
The use of citizen data for purposes other than ones, for which an explicit consent has been
given.
Quality/ Accuracy of information
Inaccurate or incomplete data can lead to erroneous or biased decisions, especially in sensitive
settings.
Transparency, Accountability Who is responsible or liable for AI making a bad decision (ethically, legally or otherwise)?
Information ownership Use of individual’s personal data for corporate benefit
Alignment of values
What values should be programmed into the AI making complex data-based decisions?
The conflict between the values of the government and citizens: between individual and public
good
Cost Direct and indirect costs of implementing AI: unemployment., disruption.
gov30.eu 21

Conclusions
 Deploying disruptive technologies in public services
brings new ethical challenges that need to be addressed
by the researchers and practitioners of e-government.
 Privacy and value alignment are major concerns both
about the AI and Big Data technologies.
 Data ownership, accuracy and use are concerns about
Big Data and accountability, inclusivity and costs
regarding AI.
 The need for legal frameworks and regulation of the use
of disruptive technologies arises in both, AI and Big Data
ethical discussions.
gov30.eu 22

Questions:
What are the Ethical responsibilities of
 Governments
 Businesses
 Citizens ?
gov30.eu 23

Addressing the issues?
 Transparency and accountability
 Algorithmic audit
 Analysis of algorithm design, data policies
 Code review, output testing
 Scrutiny of data used for training
 Ethical code of conduct
 For AI-systems developers
 For AI as independent actors
gov30.eu 24

Thank you for your attention!
Σας ευχαριστώ για την προσοχή σας!

Ethical Issues on eGovernment 3.0: Big Data and AI

In this document

More Related Content

What's hot

Similar to Ethical Issues on eGovernment 3.0: Big Data and AI

More from Samos2019Summit

Recently uploaded

Ethical Issues on eGovernment 3.0: Big Data and AI