IEEE Robotics & Automation Magazine - June 2019 - 73

One particular challenge is that, as HRI research has
shown, users intuitively relate to robots as if they were living
beings [12]. The information-processing capacities of such
robots may not be apparent to users. In particular, it may not
be evident to users that their data may be collected and analyzed. Accordingly, without prompting, users may not expect
that such information processing can raise concerns and
require explanation.
Ethics of Transparency
The ethical need for transparency can be understood as closely linked to the value of human autonomy. Autonomy
requires that users have the opportunity to interact with their
environment on their own terms. Transparency gives users of
technologies an understanding of what will be happening
with their data; having such information facilitates an
informed-consent process that allows them to make meaningful choices about their use of these technologies. According to Beauchamp and Childress [3], several elements of
consent need to be realized for an autonomy-respecting
informed-consent process. 1) Users must not be coerced into
consenting; 2) they must have the capacity to consent;
3) information that is relevant to understanding the nature
and potential impact of the technology, including practical
implications, risks, costs, and benefits, needs to be disclosed
understandably to the user; 4) users must be given opportunities to achieve understanding; and 5) users have to authorize
interventions actively.
Given the complexity and opacity of information processing in information technologies, achieving meaningful
informed consent to information technologies is challenging.
Achieving consent merely through notice and consent, via the
simple acceptance of terms and conditions for general processing purposes that are designed to meet the legal minimum of disclosure, is ethically insufficient according to these
criteria. Not only is there often little choice available to users;
frequently, such terms and conditions remain vague and
unclear on those aspects that users would consider essential
to know while providing lengthy and detailed information
that is not designed to enable users' engagement. In short,
consent procedures often do not give users understandable
information on what the system is doing and how it may
potentially affect them that is at an appropriate level of specificity to be informative without being overly demanding.
Demands for transparency need to take into account what
information would be of value and interest to potential users
to underpin meaningful decision-making and help them
engage with that information. Importantly, a lack of transparency may affect the perceived trustworthiness of those
responsible for the provision of such information. While the
ethical literature on transparency emphasizes the complexity
of potential positive and negative effects of transparency [10],
[18], it is generally acknowledged that transparency conveys
at least the willingness to be open to scrutiny and to be held
accountable. It indicates trustworthiness to potential users,
even if, in practice, transparency may not always result in

increased accountability or more significant experience of
trust among recipients of transparency information [18].
Transparency and the Law
Although transparency can be conceived from an ethical
point of view as a way to achieve autonomy via informed consent and to convey trustworthiness, the legal field has specified the conditions for how to obtain consent. It has defined
how data controllers have to inform data subjects about the
processing of personal data in a transparent manner. Today,
transparency is a core principle enshrined in Article 5 (1)(a)
of the GDPR, which states that personal data must be "processed lawfully, fairly and in a transparent manner in relation
to the data subject."
Not only has transparency become a principle of data
protection, but the term has also been specified within the
GDPR. One aspect of transparency is the broadened information duties of data controllers (defined in Articles 13 and
14 of the GDPR). Data controllers must, at the least, inform
data subjects who they are; what quantity and quality of personal data they process; and when, for how long, why, and
for what purposes they handle the data. Recitals 39 and 58
of the GDPR provide some guidance on how to implement
transparency in different systems. In particular, data controllers must inform their customers about their data processing practices through concise, easily accessible, easy to
understand, clear, and plain language (and, where appropriate, with visualization). The information must be provided
in writing or, where necessary, by electronic means, and the
information must come in an intelligible and easily accessible form (in particular, when data controllers target children; see Article 12 of the GDPR).
Apart from prospective transparency, where a data subject
is informed about the data processing beforehand, transparency requires retrospective transparency, meaning the ability
to follow the data processing step-by-step, for audit purposes,
for example [19]. Article 22 of the GDPR gives data subjects
the right not to be subject to a decision that significantly
affects them based solely on automated processing. Moreover,
Recital 71 of the GDPR gives the subject the right to obtain
human intervention, express his or her point of view, and
obtain an explanation of the decision.
Although the GDPR contains specific passages that
explain how complex data collection and transformation
processes should be made accessible to data subjects
(Articles 12-22, 34 GDPR), these descriptions are still
subject to interpretation. Additional specification and evidence on effective strategies are required to support engineers and HRI experts in realizing transparency requirements
from both the legal and ethical point of view. In particular,
improvements are necessary to the current practice of
reducing consent to the presentation of complex information to users followed by a simple tick box. New strategies
are needed concerning adaptation to users' information
needs and differentiated preferences, to allow the transparency requirement to contribute to facilitating meaningful
JUNE 2019

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

73
IEEE Robotics & Automation Magazine - June 2019

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2019
