De Fine Licht (2020) – Artificial intelligence, transparency, and public decisionmaking Flashcards
Abstract
The increasing use of Artificial Intelligence (AI) for making decisions in public affairs has sparked a lively debate on the benefits and potential harms of self-learning technologies, ranging from the hopes of fully informed and objectively taken decisions to fear for the destruction of mankind. To prevent the negative outcomes and to achieve accountable systems, many have argued that we need to open up the “black box” of AI decision-making and make it more transparent. Whereas this debate has primarily focused on how transparency can secure high-quality, fair, and reliable decisions, far less attention has been devoted to the role of transparency when it comes to how the general public come to perceive AI decision-making as legitimate and worthy of acceptance. Since relying on coercion is not only normatively problematic but also costly and highly inefficient, perceived legitimacy is fundamental to the democratic system. This paper discusses how transparency in and about AI decision-making can affect the public’s perception of the legitimacy of decisions and decision-makers and produce a framework for analyzing these questions. We argue that a limited form of transparency that focuses on providing justifications for decisions has
the potential to provide sufficient ground for perceived legitimacy without producing the harms full transparency would bring.
Opening the ‘black box’ of AI decision-making
ill make it easier for us to understand
(interpret) the functioning of the AI as well as possible to receive explanations for
individual decisions
How does an organization or sate of affairs has become transparent?
When an actor (A) has made its workings and/or performances available (or more available) (B) to another actor (C)
-This can be done through various means (M)
A distinction can be made between transparency that:
- Informs C (e.g., the public) about final decisions or policies
- About the process resulting in the decisions (transparency in process)
- About the reasons on which the decision is based (transparency in rational
The article proposes three different phases
- The goal-setting phase
- Coding phase
- Implementation phase
- The goal-setting phase
Decision-makers decide on the goals of the AI, how they should be weighed against each other when in conflict, and the features and data available to draw inferences from
- Coding phase
- The AI is developed and worked on to ensure it meets the necessary standards
- A point of introduction for problems related to bugs and biases
- It is discussed what the accuracy rates are, what they should be, how these and other
performance metrics differ, how they should be allowed to differ across different
subpopulations (when deciding about groups or i
- Implementation phase
- AI is applied in the public decision-making processes, and the results produced by the AI
are used in actual decision-making - In this phase, researchers often refer to transparency
Transparency: why more is not necessarily better
ransparency: why more is not necessarily better
* An argument for rendering the decision-making process in phases 1 till 3 fully transparent
is that transparency makes decision-makers aware of the public eye, thereby making
them aware of their responsibility to work toward the public good rather than in their own
self-interests
- Transparency in a process can lead to positive effects, but it is not the case in relation
to AI in public decision-making à the public needs a chance to actually understand
the code, and the code would therefore have to be much simpler than it is in today’s
systems
- It is better to be transparent in phase 1 and 2 à the transparency about a process
imposes incentives on the decision-makers to look credible in the eyes of the
observers during their deliberations, but this is not positive in relation to the prospect
of reaching a good decision
* Another argument in favor of full transparency à it should increase public understanding
of decisions and decision-making processes, thus making the public more confident
about decision-makers
- Information-overload/transparency paradox à if we received all of the information
from phase 1 and 2, we would have a lot of information to assimilate, and this
problem becomes even worse in phase 3
* The third argument in favor of full transparency à it increased perceived legitimacy
because it induces a feeling of control among the public
- Transparency allows the principal (the public) to overcome the information asymmetry
regarding the agent’s (representative’s) workings, thereby leading to renewed
instructions about what to do or even removal from office
* The final argument in favor of full transparency à transparency generates positive results
regarding perceived legitimacy, as the public will perceive the decision-making processes
to be fair, and this view will also affect their evaluations of the decisions and decisionmakers
- The public may become disappointed when they realize that decision-making
processes are, characterized by a process of ‘muddling through’ rather than a rational
process of identifying
The benefits of a justifications approach to AI transparency
- A policy of justifications of decisions will inform the public of what the decision is, on
which grounds it has been made, and identify who the responsible actor is - Decision-makers should provide favorable reasons for the goals and priorities they have
established for the AI and ensure that these reasons are made available to the public in a
way they can understand - Principal-agent terminology à the agent (people) will know who to hold accountable and
for what, and this appears true for all phases - The decisions that the decision-makers make will probably be or higher quality because
of what the decision-makers need to do and what they can do before facing the public
Gedownload door: jjverhelst | jjverhelst@outlook.be
Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar.
€ 912 per jaar
extra verdienen?
Stuvia - Koop en Verkoop de Beste Samenvattingen - Obvious objection à focusing on the provision of justifications might increase the
likelihood of decision-makers engaging in a post-decision construction of arguments
designed to look better than they actually are - Transparency is generally associated with the ‘myth of hidden politics’ à public
perception that the actual decision-making is something that takes place in smoke filled
rooms or private spaces that are hidden from the public arena - Also, there are reasons to believe that transparency in rationale could yield as much or
even higher degrees of understanding - In phase 1 and 2, when decision-makers are transparent about the decision and the
reasons for taking it, the public should have greater understanding of why these
decisions were made, since the information about the decision will be more
condensed and thus more accessible - Opening the black box in phase 3 in the sense that the AI gives explanations for its
decisions rather than describing the whole process will also lead to greater
understanding for the public - The public will likely view transparency in rationale as a fair way of dealing with these
issues
Conclusion of the article
First, there is a need for a thorough analysis of what reasons should be normatively acceptable to use when publicly justifying decisions in a democratic setting
- Second, there is a need for more empirical research regarding how justifications should be designed and presented to gain public acceptance
- As argued by McGraw et al. (1995), several conditions must be met for an explanation/justification to have the intended effect: It must be exposed to the intended audience, the audience must pay attention to it, the audience must comprehend it, and the audience must accept the explanation/justification as legitimate
- Third, there is a need to develop an empirically grounded theory for how a policy of
justifications should be designed in practice to satisfy the demands of public insight and efficiency with regard to decision-making - Fourth, we need to evaluate how decisions and decision justifications are perceived by
the public, depending on whether they are being made by human beings or AIs
