Lecture 4 Flashcards
Principles of Bureaucracy & citizen-state interaction
- Use of AI may change citizen-state interactions on a fundamental level
- The workings of bureaucracy is co-created by individuals, organizations, and institutions
- These agents‘ relationships are governed by rules and public values
What are the rules of Principles of Bureaucracy & citizen-state interaction?
- The essential strength and objective of bureaucracies is the non-
discriminatory implementation of policy - …without any preferential treatment (Weber 1922)
- Ideal-type bureaucracies are populated with
“bureaucrats [who] are responsible for following rules with regard to their office
with dedication and integrity and for avoiding arbitrary action and action based
on personal likes and dislikes.” (Olsen 2005)
Why may AI change citizen-state
interactions on a fundamental level
To understand this, contexts and
administrative traditions are very
important
- How government works varies across
different socio-economic and cultural
settings - Administrative traditions define the
relationship between agents by setting
principles and hierarchies of public
values
Why may AI change citizen-state interactions on a fundamental level?
AI governs in subtle yet fundamental ways the way we live and are transforming our
societies; AI blurs value hierarchies and moral attribution.
* The use of AI disrupts this configuration by breaking the Responsibility
Accountability logic, hence undermining fundamental principles of modern
bureaucracies
* “Accountability refers to being answerable to somebody else, being obligated to
explain and justify action and inaction”;
* “accountability is defined as “a relationship between an actor and a forum, in which
the actor has the obligation to explain and justify their conduct, the forum can pose
questions and pass judgement, and the actor might face consequences.”
What are consequences & risks of undermining the accountability logic by AI:
Algorithms are not neutral and introduce new “machine biases” to human cognition
- Low-stakes AI applications: e.g.
- text prediction on mobile phone apps
- speech-to-speech translation
- tweet bots
- High-stakes AI applications, e.g.:
- crime prediction & policing
- semi-autonomous driving
- individual corruption score prediction
- medical assessment
- Fundamental problem: Deep learning tools are inherently
opaque - Most algorithms are proprietary
- AI is complex and rapid
- Currently: Value trade-offs are implicit
undermining democratic principles
Strong call for regulatory efforts!
Echoed by recent call for alignment:
Can AI be aligned with our
values?
- Public values govern various
administrative relationships, what
values does AI have? - AI research and development should
be refocused on making today’s
powerful, state-of-the-art systems
more accurate, safe, interpretable,
transparent, robust, aligned,
trustworthy, and loyal. - AI research and
development should be
refocused on making
today’s powerful, state-
of-the-art systems more
accurate, safe,
interpretable,
transparent, robust,
aligned, trustworthy,
and loyal.”
What are Common arguments for the use of AI in administration?
- Using algorithms to replace or assist decision-making is more cost efficient.
- Efficiency gains despite the age of austerity
- Human cognition is inherently biased (e.g. Kahneman & Tversky 1979)
- AI (presumably) lacks malintent (Gamez et al. 2020)
Some people havemore trust in the moral quality of machine-based decision-making
compared to human decision-making - Three dimensions of trust (Mayer et al. 1995): ability, benevolence,
integrity - AI may boost digitalization of public administration, resulting in
- Improved accessibility
- Higher transparency
- Procedural accountability through traceability
- AI may help un-bias human decision making
- in coordination tasks in network governance
- or in Human Resource Management (Maasland & Weißmüller 2021,
Keppler 2022)
What is the defintionof corruption according this course?
abuse of entrusted power for private gains
How to strengthen bottom-up AI-ACT
