Minds and Machines (2021) 31:563–593

https://doi.org/10.1007/s11023-021-09573-8

GENERAL ARTICLE

Linking Human And Machine Behavior: A New Approach

to Evaluate Training Data Quality for Beneficial Machine
Learning

Thilo Hagendorff1 

Received: 28 July 2020 / Accepted: 9 September 2021 / Published online: 26 September 2021
© The Author(s) 2021

Abstract
Machine behavior that is based on learning algorithms can be significantly influ-
enced by the exposure to data of different qualities. Up to now, those qualities are
solely measured in technical terms, but not in ethical ones, despite the significant
role of training and annotation data in supervised machine learning. This is the first
study to fill this gap by describing new dimensions of data quality for supervised
machine learning applications. Based on the rationale that different social and psy-
chological backgrounds of individuals correlate in practice with different modes of
human–computer-interaction, the paper describes from an ethical perspective how
varying qualities of behavioral data that individuals leave behind while using digi-
tal technologies have socially relevant ramification for the development of machine
learning applications. The specific objective of this study is to describe how training
data can be selected according to ethical assessments of the behavior it originates
from, establishing an innovative filter regime to transition from the big data ration-
ale n = all to a more selective way of processing data for training sets in machine
learning. The overarching aim of this research is to promote methods for achieving
beneficial machine learning applications that could be widely useful for industry as
well as academia.

Keywords  Artificial intelligence · Machine learning · Machine behavior ·

Technology ethics · Training data · Data quality

* Thilo Hagendorff
[email protected]
1
Cluster of Excellence “Machine Learning ‑ New Perspectives for Science”, University
of Tuebingen, Tübingen, Germany

13
Vol.:(0123456789)
564
T. Hagendorff

1 Introduction

When developing learning software, practitioners have additional ethical respon-

sibilities beyond those of standard, non-learning software (Wolf et  al., 2017).
These responsibilities originate in the careful selection of inputs that build the
very basis for the computational learning process itself. In a recent paper on
machine behavior, Rahwan et al. stress that machine learning applications “can-
not be fully understood without the integrated study of algorithms and the social
environments in which algorithms operate” (Rahwan et  al., 2019, p.  477). With
regard to supervised machine learning, meaning artificial neural networks, sup-
port vector machines, naive Bayes classifiers, regression algorithms etc., those
“social environments” can, among others, be understood as different training
stimuli that shape the behavior of a machine. Machine behavior can be seen in
analogy to the behavior of biological agents as an observable response to (inter-
nal or) external stimuli. Training data fed into supervised machine learning appli-
cations reflect, in case it is about behavioral data, people’s (e.g., discriminative)
behavior, so people’s behavior has an indirect influence on machine (discrimina-
tive) behavior (Barocas & Selbst, 2016). This influence cannot be described as
a direct relationship, meaning as an equivalence between people’s behavior and
machine behavior. However, particular traits of learning machines, that is how
they solve for instance classification, prediction, or generation tasks, can origi-
nate in similar features that are part of behavioral data sets. Thus, when technol-
ogy ethicists talk about “moral machines” (Wallach & Allen, 2009) in the con-
text of machine learning applications, one also has to ask for “moral people” and
“moral people’s data”, to put it simply. Of course, these “moral machines” are
also the result of engineering or design choices, they are dependent on the selec-
tion of hyperparameters or specific wirings of artificial neural networks, and the
like. But in general, today’s machine learning techniques are dependent on human
participation. In many cases, they harness human behavior that is digitized by
various tracking methods. These machine learning methods do not create intel-
ligence, but, taking up the figurative words of Mühlhoff, “capture” (2019, 1873)
it by tracking human cognitive and behavioral abilities. Without the empirical
aggregation of recordings of human behavior, many parts of machine learning
would not work. An extensive infrastructure for “extracting” (Crawford, 2021,
p.  15) valuable personal data or “capturing” human behavior in distributed net-
works via user-generated content, expressed or implicit relations between people,
as well as behavioral traces (Olteanu et al., 2019) builds the bedrock for a compu-
tational capacity called “artificial intelligence” (Mühlhoff, 2019).
Here, I want to ask whether there are differences in the “quality” of human
participation in artificial intelligence. To do this, one must further answer the
question about what constitutes “good” influences or “good” behavioral datasets
for supervised machine learning applications. In order to accomplish this, I will
focus on identifying data sources reflecting behavior that is ethically sound, which
in turn can be identified via scrutinizing particular states and traits of an individ-
ual that are to be described in more detail. With the help of a matrix of different

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 565

evaluation frameworks, a normative evaluation of different data sources can take

place. To the best of my knowledge, such an approach has not yet been enlarged
upon in the computer sciences. Hitherto, normatively oriented machine learning
research is mainly concerned with fairness (Kearns & Roth, 2020) or preventing
discrimination (Hagendorff, 2019c), robustness (Amodei et al. 2017), explainabil-
ity (Mittelstadt et al., 2019), or preserving privacy (Dwork et al. 2006). Besides
that, especially in the field of supervised machine learning, the question of what
characterizes—from an ethical perspective—good data contexts remains largely
unanswered. This is crucial, since morally sound machine learning applications
are in many regards only as sophisticated as their “environmental influences” or
training and supervision stimuli. This can be illustrated well with the example of
labels in medical machine learning applications. When X-ray images or other out-
comes of medical imaging techniques are annotated, this is not done via crowd-
or outsourcing labeling procedures, but via expert labeling (Irvin et  al., 2019).
Why? Because experts are versed enough to provide the right decision-making
routines that can help medical machine learning to work properly. So why not
extend this tenet of selecting for expert data or for versed individuals and captur-
ing only their cognitive and behavioral abilities in other areas of machine learn-
ing? This paper will elaborate on this idea in detail.
Fruitful research can emerge when the social sciences are combined with
machine learning research, so that not only ethics, but also technology develop-
ment can be advanced. Most research works in this area provide critique rather
than engage constructively by creating positive ideas and visions on how to use
machine learning technologies for the common good. This paper stands in line
with and continues the “good data project” (Daly et al. 2019a), promoting tangi-
ble good and ethical data practices and frameworks instead of mainly criticizing
what goes wrong with machine learning and big data applications. The follow-
ing chapters elaborate on that in more detail. Chapter  2.1 describes the present
approach which is to use as much behavioral data as possible for machine learn-
ing development. Although criteria for data quality exist to prefilter training stim-
uli, these criteria are solely oriented along technical dimensions, not ethical ones,
as depicted in chapter 2.2. Chapter 3.1 then describes how human behavior is nor-
matively classified in sociology and psychology, while chapter 3.2 describes how
tracking technologies can be inspired by those classifications in order to single
out datasets from certain subpopulations that are deemed to be the most compe-
tent or morally versed group for a particular task. Chapter 4 then investigates five
particular applications of machine learning, namely autonomous cars, language
generation, search engine ranking algorithms, social media filtering systems, and
e-commerce recommendation systems, that can be made more beneficial by fol-
lowing the presented ideas. To that end, machine behavior objectives, behavioral
data sources, tracked states and traits to assess the quality of those data, as well
as quality training stimuli of each of the example applications are to be described.
Subsequently, chapter  5 covers some points of discussion and responds to them
defending the presented approach for beneficial machine learning. Finally, chap-
ter 6 concludes and sums up the paper’s arguments.

13
566
T. Hagendorff

2 More Data are not Always Better: Defining New Dimensions

of Data Quality

2.1 The Idea of n = all and its Shortcomings

Before deliberating on data quality dimensions, I want to recapitulate the tenets

of big data. Big data meant the emancipation from small data studies, a paradigm
scientific knowledge discovery relied on for hundreds of years. Big data lead to
the success of today’s machine learning systems, which are heralded as the new
gold standard of knowledge discovery since they are necessary to understand
increasingly complex collections of data, especially in the sciences (Jordan &
Mitchell, 2015; Mjolsness & DeCoste, 2001). Broadly speaking, this trend caused
some kind of “amnesia” on the value of small data (Kitchin & Lauriault, 2015).
While small data are of narrow variety, have a limited volume, are generated to
answer specific questions, and produced in controlled ways, big data are the exact
opposite. The latter are large in volume since they are generated continuously as
a by-product of digital technologies. As stated many times, big data strives to
be exhaustive in scope, or, in other words, it follows the ideology of n = all. The
formula n = all encapsulates the idea that “more trumps better” (Mayer-Schön-
berger & Cukier, 2013, p.  13; Perrons & McAuley, 2015). Hence, big data are
often indifferent towards predefined, specific queries or areas of interest in the
context of which one wants to gather insights. Queries often repurpose data to
gain insights into phenomena that have no or only indirect linkage to the original
context of the data acquisition.
Machine learning techniques allow probabilistic inferences on unknown fea-
tures. This is why current machine learning applications work under the motto
“the more data they have, the better they get” (Domingos, 2015, p. xi). But when
speaking about behavioral data, this claim may not be true. It seems that learn-
ing applications do not have to be programmed, they program themselves. But
when they program themselves while being fed with as much behavioral data as
possible in order to aim at higher grades of accuracy, they also become indif-
ferent with respect to the orientation towards certain moral values. The ideol-
ogy of n = all leads to technical systems that utilize an endless stream of choices
made by humans interacting with online platforms and digital devices—a practice
once called “laissez-faire data collection” (Jo and Gebru 2019), narrowing down
everything towards scores which represent averages of whole populations. But
instead of simply recognizing patterns within datasets of a whole population, one
could single out datasets—and hence training stimuli—from a certain subpopula-
tion, namely the most competent, eligible, or morally versed one for a respective
task, and find patterns only within this data context. Subsequently, only those pat-
terns provide the basis for the generalization ability of a given model or learning
algorithm. By diversifying or sampling various data contexts in the larger frame
of big data, one can reintroduce ideas connected to the concept of small data in
the current situation of an abundance of data. This abundance is so prevalent
that measures to tailor training data sets for machine learning applications with

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 567

respect to certain fractions of large data sets does not necessarily mean to signifi-
cantly restrict technical capabilities of these applications. Notwithstanding this,
advanced machine learning methods are able to learn from small datasets via data
augmentation, can generate synthetic data via GANs or variational autoencoders
to artificially increase the amount of training stimuli, use transfer learning to use
knowledge from an already learned task, utilize few shot learning mechanisms,
etc. (Shorten & Khoshgoftaar, 2019; Zhuang et al. 2020; Mayer et al., 2018). In
short, learning algorithms are not reliant on data abundance. This allows for a
transition from n = all to n = X, where X ⊊ all. But the decisive question is what
subgroups or fractions to choose, what data nuances to accentuate. In this con-
text, the claim that “more trumps better” is transposed to “better trumps more”.
But what is better data? To answer this question, one has to look at data quality.

2.2 Data quality Dimensions and Ethics

A common saying in computer sciences is “garbage in, garbage out”, referring to the
importance of quality data in data-intensive applications like supervised machine
learning. Surveys found that data quality attributes comprise literally hundreds of
variables (Wang & Strong, 1996). Nevertheless, discourses on data quality define it
in terms of its suitability for a business purpose and decision-making efficiency in
companies (Samitsch, 2015), and are solely focused on particular technical dimen-
sions like data cleanliness (how many errors do data sets contain?), data complete-
ness (how exhaustive for a particular task are data sets?), data objectivity (what
biases do data sets contain?), data consistency and reliability (how many discrepan-
cies are contained in data sets?), data timeliness (how current are the data?), data
veracity and exactitude (how accurate and precise are information in data sets?),
data interpretability (how readable are data sets?), data cost-effectiveness (what are
the costs of data collections?), and the like (Gudivada et al., 2017). Similarly, data
quality problems are defined in terms of missing data, duplicate data, inconsistent
data formats, incorrect values, spelling errors, etc. (Woodall et  al., 2014). Current
approaches to improve datasheets for datasets also do not include aspects that go
beyond technical and organizational items like “Who funded the creation of the
datasets?”, “Are there any errors, sources of noise, or redundancies in the dataset?”,
“Will the datasets be updated?”, and so on (Gebru et al., 2018). All those questions
and differentiations make perfect sense when assessing data sets that do not contain
data that relate to human behavior. But in case data sets relate to it, the discourse on
data quality has to be extended.
Besides the data quality discourse, a further discourse addresses the construction
of digital persons via data traces from volatile and non-volatile data acquisitions,
sensors of all kinds, surveillance measures, social media platforms, and the like.
Personal data from different sources and domains are linked together in a form of
a “dense rhizomatic assemblage” (Kitchin & Dodge, 2011, p. 90). Terms like “data
subjects”, “data derivatives” (Amoore, 2011), “data double” (Los, 2006), “shadow
order” (Bogard, 1996), “digital persona” (Clarke, 1994), “dividuals” (Deleuze,
1992), or “data doubles” (Lyon, 2003) are used to describe the comprehensive

13
568
T. Hagendorff

compilation of personal data, the creation of increasingly detailed and fine-grained

digital footprints of individuals, which are then later processed in machine learn-
ing applications, which in turn have various (and in some cases negative) ramifica-
tions for society (Calvo et al., 2020; Eubanks, 2018; O’Neil, 2016). In this process,
various “filters” mediate the translation from an individual’s original behavior to
eventual computer outputs. Those filters take effect through the selection of certain
sensors, data cleansing processes, feature extraction, software libraries, data visuali-
zations, etc. This is why the literature on critical data studies claims that something
like “raw data” does not exist (Gitelman, 2013). There are, metaphorically speak-
ing, always bottlenecks, strainers, gates, intentionally or non-intentionally regulat-
ing the “permeability” for data at different stages of the computational processing
of reality. But those filters do not have an ethical dimension. They do not lead to an
ethically motivated selection and sorting out of different data contexts with varying
ethical data qualities. To define what I mean by ethical data qualities, one has to
analyze how data quality is affected by certain personality traits or modes of behav-
ior of individuals, and how those traits or states can be assessed from an ethical
point of view. Eventually, finding quality data shall not primarily serve the pursuit
of an improved marketability, but of socially accepted, beneficial machine learning
applications.

3 Human Behavior and its Digital Records

3.1 Classifying Human Behavior

Typically, behavioral data are the result of tracking online activities of all kinds,
meaning user-generated content, expressed or implicit relations between people, or
behavioral traces (Olteanu et  al., 2019). Different modes of behavior eventuate in
different data contexts. Individuals leave different data traces behind depending on
their emotional state, educational background, intelligence, wealth, age, moral matu-
rity, and the like. In order to sort those traits and to classify human behavior and
stages of development, one can draw on well-established theories in psychology and
sociology. Within the framework of these theories, the aim is to distinguish differ-
ent modes of behavior or stages of development according to empirical findings.
As a general rule, behavior or personality development is understood to be largely
a product of one’s social environments. Those environments are classified, for
instance, with the help of theories of social stratification (Bourdieu, 1984; Erikson
et al., 1979; Grusky, 2019; Schulze, 1996; Vester, 2001). A person’s milieu, mean-
ing, simplistically speaking, upper, middle, or lower classes, determines their habi-
tus, which in turn determines parts of their behavioral routines and vice versa. Indi-
viduals occupy a certain position in “social space” which is the result of a contested
distribution of resources, meaning economic, cultural, social, or symbolic capital
(Bourdieu, 1989). The position an individual occupies in social space is in large
parts “hereditary” and can be affected by social injustices. Nevertheless, the amount
of capital a person can concentrate on her- or himself has a structuring power on

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 569

many areas of life, meaning that it organizes a person’s taste, language, estate, politi-
cal orientation, or, to say it more generally, his or her dispositions.
Further, these dispositions also structure and have an impact on the way a person
uses digital technologies, and influence what kind of data are tracked by these tech-
nologies. By using terms like “media-based inequalities”, “digital divide” or “digital
inequality”, several studies show the strong influence a user’s socioeconomic status
has on media or Internet usage patterns (McCloud et  al., 2016; Zillien & Hargit-
tai, 2009; boyd 2012; Hargittai, 2008; Mossberger et  al., 2003). Individuals with
a higher socioeconomic status are more likely to engage in online activities that
enhance their social position, have status-specific interests, interact more frequently
with e.g. political or economic news or health information, have higher levels of
computer literacy, use less often chat platforms or social networking sites, and so
forth. All in all, the position of an individual in social space heavily influences his or
her ways of using digital technologies and hence the kind of behavioral data that are
digitally recorded—with the respective consequences for biases, scopes, representa-
tive statuses, or ethical quality dimensions of data sets.
While behavior is in many respects an outcome of the respective social environ-
ment, class, milieu, or social position, the same holds true for personality develop-
ment, which is widely dependent on the circumstances of socialization. Develop-
mental psychological theories either postulate that personality development follows
a continuous process or that it passes through discontinuous stages, where logical
reasoning is learned, moral senses are developed, social norms are adopted, emo-
tional intelligence is acquired, stereotypes are negotiated, role models are changed,
self-reflection is learned, values are internalized, personal crises are overcome, and
the like (Erikson, 1980; Kohlberg et al., 1983; Loevinger, 1997). In order for an indi-
vidual’s socialization to succeed, it requires, among other things, a certain range of
beneficial influences from a social environment, which can be separated from harm-
ful influences. To scrutinize these influences is the objective of developmental psy-
chology. The discipline focuses on long-term progressions with regard to the experi-
ences and the behavior of individuals in order to find patterns and regularities that
are crucial for the development of intellectually and emotionally sound and mature
individuals (Lerner 2015). A succeeding development is measured by aspects such
as problem-solving abilities, emotional intelligence, cognitive development, proso-
cial behavior, mental health, educational success, etc. As soon as such norms for a
successful personal development are defined, one can roughly differentiate between
positive and negative environmental influences. The latter can affect health, gross
and fine motor skills, socio-emotional development, the speed of information pro-
cessing, self-concepts, knowledge, or language behavior and range from alcohol to
stress during pregnancy, residential areas with high crime rates, low educational lev-
els, emotional, physical or sexual abuse, as well as a neglectful parenting style (Sul-
livan and Knutson 2000; Spera 2005).
According to theories from developmental psychology, “higher” forms of per-
sonality development lead to other behavior patterns than “lower” ones (Hart et al.,
1997; Paul B. Baltes et al. 1978; Kohlberg et al., 1983). Normally, more cognitive-
moral growth leads to more socially desirable or acceptable behavior. Philosophical
theories about ideal moral acting, ranging from Kant’s categorical imperative (Kant,

13
570
T. Hagendorff

1977), Habermas’ discursive will-formation (Habermas, 1987), or Rawls’ theory of

social contract (Rawls, 1999), imply that individuals possess fully developed cogni-
tive capacities. In this context, one can assume that personality or character develop-
ment may strive towards the target values and rationality standards of these models.
In order to measure the “proximity” of a person’s character to certain target values,
differential psychologists have developed various tools for personality assessment
to understand and predict behavior in different social contexts. Amongst personality
assessment tools are the widely used Five-Factor Model (John et al., 2008; McCrae
& John, 1992), non-scientific tests like the Myers-Briggs type indicator (Myers and
Myers 1995), or less known methods like the Multidimensional Personality Ques-
tionnaire (Tellegen & Waller, 2008), “PerformanSe” (Patel, 2006), and many more.
All these tools have specific weaknesses, they ignore the fact that personality can be
in a state of flux, and that it may be unclear what personality characteristics mean
in terms of behavioral manifestations in certain situations. But apart from that, they
more or less reliably measure traits like motivation, extraversion, emotional stability,
openness, conformism, rationality, impulsivity, dynamism, anxieties, social activity,
and the like.
The mentioned theories and tools have a tacit consensus about certain ethical
target values or “attraction poles” (Sloterdijk, 2009). This can be exemplarily elu-
cidated with regard to the Five-Factor Model (John et al., 2008). All five personal-
ity dimensions have an attraction pole, meaning that all dimensions can be spanned
between two poles whereas one pole is designated as the favored one. Typically,
more complexity in an individual’s mental and experiential life is better than less
(openness), more impulse control that facilitates goal-directed behavior is better
than less (conscientiousness), more social activity and positive emotionality is better
than less (extraversion), a more prosocial and communal orientation is better than
less (agreeableness), and more emotional stability is better than less (neuroticism).
In the background of personality assessment tools, developmental psychological, or
social milieu theories, tacit normative presuppositions exist that structure attraction
poles of all kinds. However, making these presuppositions and polarizations explicit
may not be equated with attempts to classify humans as such. The mere idea of
classifying humans provokes strong moral intuitions to refuse such practices. But
besides these moral intuitions, the application of classification or scoring systems
on humans is common industry and government practice in many countries (Engel-
mann et  al., 2019). People are classified with respect to their financial situation,
their social reputation, their risks of conducting certain actions, their personality,
etc. Moreover, they are classified along geodemographic segmentations, purchasing
histories, lifestyle types, and the like. However, the circumstance that the application
of ranking systems on people corresponds to the status quo of the digital economy
does not mean that the related practices are morally correct. Quite the opposite may
be true (Zuboff, 2015). In the context of this paper, though, I do not propose to apply
digital ranking practices to individuals as such, but to particular types of behavior
or particular personality traits. They can be measured alongside certain dimensions,
which have a more or less strong normative alignment. Hence, in the following
chapter, I want to elaborate on how personality traits and different types of human
behavior can be digitally measured and classified.

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 571

3.2 Tracking Human–Computer Interaction

Dataveillance (Clarke, 1988; van Dijck, 2014), in other words practices of recording
and analyzing digitally mediated behavior, has at least three complications or down-
sides. First, it is a morally contested practice, causing negative “chilling effects” of
all kinds (Schneier, 2015). Second, monitoring human–computer interactions or
online behavior does not yield data that corresponds to real attributes but it con-
structs them (Haggerty & Ericson, 2000). And third, one can only infer personal
attributes with the right data bases. Obviously, missing data strictly limits the scope
of information one can gain, albeit the possibility of statistically inferring informa-
tion on unknown features exists. Before these three complications or downsides are
discussed in more detail in chapter 5, it is important to see that in practice, various
tools are used to track personal traits and states (Matz & Netzer, 2017). Single- as
well as multimodal approaches that combine several psychological attribute recogni-
tion methods, and that can detect involuntary (e.g. physiological), semi-voluntary
(e.g. facial expressions), as well as voluntary (e.g. key presses) signals are used
(D’Mello & Kory, 2015; Zeng et al., 2009).
Some instances are listed hereafter: By analyzing clickstreams, browsing histo-
ries, or search queries, inferences on users’ demographic information can be made
(Acar et al., 2014; Bi et al., 2013; Hu et al., 2007). Affective computing serves to
detect emotions, mostly through text, voice, face, or posture processing techniques
(Picard, 1997). Written text can be investigated in order to detect mental illnesses, to
conduct sentiment or personality analysis via differential language analysis, natural
language processing, and other machine learning methods (Guntuku et  al., 2017;
Pang & Lee, 2008; Schwartz et  al., 2013). Moreover, digital images, for instance
social media profile pictures, can also be used to reveal personality attributes (Sega-
lin et  al., 2017). Various sensors—especially the ones in smartphones and other
wearable devices—are used to track physiological signals, movements, activity lev-
els, mobility patterns, face-to-face encounters, and the like in order to infer internal
states and personal attributes (Harari et al., 2017; Kwapisz et al., 2011). User input
via display touching behavior, mouse movements, or keyboard strokes can also be
used to infer personality traits (Khan et al., 2008). Many other applications could be
added.
All in all, the baseline is clear. Tracking technologies for digitally mediated
behavior can in many cases successfully measure a broad spectrum of psychological
traits, affective states, and personal attributes. Many tracking applications specifi-
cally aim at measuring the six basic emotions (sadness, fear, anger, disgust, joy, sur-
prise), while in practice, though, these basic emotions can be observed only seldom,
instead engagement, confusion, boredom, curiosity, frustration, and happiness are
more frequent affective states in human–computer-interaction (D’Mello, 2013). But
tracking technologies can also measure more complex attributes like age, gender,
sexual orientation, occupation, mean income, ethnicity, religious views, political
attitudes, personality traits, intelligence, pregnancies, use of addictive substances,
job performance, parental separation, and many more (Kosinski et al., 2013, 2014).
Here, the connection to psychological as well as sociological approaches to classify
human behavior can be made. While many classical theories from the humanities

13
572
T. Hagendorff

approach social structures and individual-related traits and distinctions on a very

high and abstract level, digital behavior tracking technologies only capture “micro-
scopic” behavior traces in data sets. But the former and the latter can be combined in
order to transition from the n = all ideology to n = X, where X ⊊ all, which means to
single out datasets from certain subpopulations that are deemed to be the most com-
petent or morally versed group for a particular task.
Tracking technologies, even if they are themselves ethically contested, can be the
bedrock for an ethically motivated selection of different data contexts with varying
ethical data qualities that can then lead to beneficial machine learning applications
and machine behavior. When recognizing that a person’s dispositions structure and
influence the way she or he uses digital technologies, then methods to detect digi-
tally mediated behavior can, in turn, infer those dispositions when analyzing data
traces. That means that “higher” states of personality or moral development, socially
desirable or acceptable behavior, distinct cognitive abilities, emotional stability,
rationality standards or, in general, the “proximity” of a person’s dispositions to cer-
tain socially accepted and ethically defined target values can be measured. Problem-
solving abilities, emotional intelligence, cognitive development, prosocial behavior,
educational status, mental health—all those assessment dimensions have attraction
poles that are used in many social contexts to rank human behavior and to assess
whether particular individuals can be put in charge, or are competent enough or eli-
gible for certain tasks. This principle is, at least when adopting a meritocratic per-
spective (Young, 1994), effective in many social institutions. From here on, I want to
make the transition to prefiltering training stimuli for machine learning applications
according to certain individual attributes, states, and traits.

4 Beneficial Machine Learning: Putting the Approach into Practice

In the context of (supervised) machine learning development, there are three ways in
which hereditary and “environmental” information can be inscribed into algorithms
(Rahwan et  al., 2019, p.  480): they can be incorporated into applications by pro-
grammers making design choices in algorithms (Brey, 2010; Friedman and Nissen-
baum 1996), by particular training stimuli (i.e. data), or by a machine’s own “expe-
riences”. Taking up the humanistic differentiation between hereditary factors and
environmental influences that shape an individual, one can stress that machine learn-
ing applications also combine both; the former through algorithm design, and the
latter through training stimuli, where both factors interfere with each other. Train-
ing stimuli, in other words a set of examples used for learning, are used to fit and
tune the architecture, parameters, or weights of a classifier. Training data sets are
supposed to allow artificial neural networks to generalize from the sample of the
training data set to potentially every other case, meaning that the network has the
best possible performance on any new data. In this context, training stimuli must
be distinguished from validation and test or holdout sets, where the former serve
the purpose of tuning the architecture of a classifier, and the latter of measuring the
performance of a trained classifier. Supervised machine learning predicts a categori-
cal or continuous value Y in the form of target variables or labels given an input X

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 573

in the form of a set of variables by training a function F, where F(X) = Y (Ghani &
Schierholz, 2017). Here, the set of variables can represent behavioral data, but one
also has to keep in mind that the same holds true for labels. Labels are most often
the result of manual clickwork (Irani, 2016), but they can have the form of genuine
behavioral data, too. For instance, this is the case when labels for video data of an
autonomous vehicle’s surroundings are generated by capturing the driver’s behavior
(Tsutsui et al., 2018). This way, human behavior also becomes machine behavior via
labels, not just via training data itself.
Concepts that follow the idea of an ethically motivated selection or limitation
of particular training data or labels in order to influence the process of developing
certain machine behaviors are non-existent up to this point. One single exception
is Davidow and Malone’s cursory concept of “starving AI” or of “putting artificial
intelligence on a data diet” (Davidow & Malone, 2020). The idea is to ensure trust-
worthy artificial intelligence not by controlling it, but by putting it in “virtual pris-
ons”, meaning that the applications are disallowed to use whatever training stimuli
they can get for learning. The authors follow a rather metaphorical approach, but,
in a nutshell, they rudimentarily capture an idea similar to the one presented in this
paper, namely developing beneficial machine learning applications by filtering train-
ing stimuli according to ethical considerations. My line of argument starts at the
assumption that a person’s social background, educational level, personality, intel-
ligence, etc. shape his or her way of using digital devices. Moreover, these devices
are equipped with sophisticated tracking technologies that can in many cases accu-
rately measure and infer the user’s personal attributes, traits, or states. Depending on
these measurements, data traces the respective user produces, that is data traces that
provided the basis to the measurements itself as well as data traces that are situated
in the same context as the measurement, are assessed from an ethical perspective.
Thereafter, this assessment enables data scientists to relinquish the idea of using as
much relevant data as possible that represent averages of whole populations. Instead,
they single out quality data that are representing behavior of specific subpopulations
which are deemed to be especially competent, eligible, or morally versed for a par-
ticular task. This limitation, that stands in contrast to the credo that the bigger the
data the better the machine learning models, serves to tailor training data in a way
that machine behavior can be steered into a direction that promotes its beneficence.

4.1 Concrete use Cases

To further elaborate on that, I want to sketch out use cases that exemplarily illustrate
the process that is outlined above. For that end, I investigated five machine learn-
ing driven applications and demonstrate how beneficial machine behavior objectives
can be achieved by selecting certain quality data contexts for model training. The
investigation shall delineate how the paper’s ideas can be put from theory into prac-
tice. I will focus only on applications that are widely used, like e-commerce recom-
mendation systems, search engine ranking algorithms, or autopilots in self-driving
cars, and describe how these applications can be amended by following a stringent
approach for quality training data selection along particular ethical considerations.

13
574
T. Hagendorff

4.1.1 Self‑Driving Cars

4.1.1.1  Machine Behavior Objectives  Autonomous vehicles are supposed to guaran-

tee as much safety as possible (Koopman & Wagner, 2017). Avoiding crashes with
self-driving cars (Xu et al., 2019) is paramount to advance their deployment. Achiev-
ing this goal has many dimensions, but it certainly encompasses safe machine behav-
ior, meaning a car complies with safe overtaking maneuver, following, emergency
stop, cornering, or line choice rules.

4.1.1.2  Data Sources  Autonomous cars must infer from past traffic situations to
new ones. Thus, training data, meaning video recordings and further sensor data
of all kind representing countless hours of driving, as well as annotations for these
data are of utmost importance. Because it is rather expensive and in some cases
notoriously difficult to acquire enough annotation data, in some autonomous vehi-
cles, label collection happens via measuring behavioral cues from human driving
behavior, e.g. acceleration, deceleration, steering, etc., in manual mode or during
autopilot disagreement (Eady, 2019). The labels are then linked to the respective
footage of the vehicle’s surroundings. Additionally, data traces from actual driving
can be combined with customer data as well as further behavioral data from third
party organizations like data-brokers.

4.1.1.3  Tracked States and Traits  There are certain individual characteristics like
gender, age, driving experience, distraction, attention, reaction time, visual func-
tion, sensation seeking, impulsivity, etc. that predict risky driving behavior (Anstey
et al., 2005; Fergusson et al., 2003; Wayne & Miller, 2018). According to accident
statistics and empirical investigations, individuals who cause fatal as well as non-
fatal car crashes tend to be male, of young age, have high levels of aggressiveness,
sensation seeking, and impulsivity as well as some other traits like lower levels
of income, poor mental health status, higher levels of neuroticism, possibly raised
blood alcohol concentration, lower driving experience, and show various forms
of antisocial behavior or higher levels of social deviance (Abdoli et  al., 2015;
Čubranić-Dobrodolac et al., 2017; Hyman, 1968; Vaughn et al., 2011; Wang et al.,
2019; West & Hall, 1997). Many, if not all of these traits can be digitally detected
at some degree of accuracy. Those characteristics as well as additional cues like
engine speed, pedal pressure, improper following, speaker volume, driver body
posture, gestures, head movement, verbal outbursts, etc. can be digitally tracked in
order to predict a driver’s safety level (van Ly et al., 2013).

4.1.1.4  Quality Data Contexts  As soon as the above-mentioned traits are digitally
tracked and recorded, the driving behavior data that is related to the respective
driver can be excluded or downgraded from the data set that is used to train the
models that determine the machine behavior during autopilot. Traffic psycholo-
gists can help machine learning practitioners to further establish tools to clas-
sify data that represent decent driving behavior. In short, quality data contexts
arise from drivers who possess decent driving experience, have a good reaction

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 575

time, tend to be female, have low levels of aggressiveness, sensation seeking and
impulsivity, show active head movement in traffic, distinguish oneself in few or no
verbal outbursts, proper following behavior, or modest acceleration behavior, to
name just a few attributes.

4.1.2 Language Generation

4.1.2.1  Machine Behavior Objectives Chatbots as well as speech assistants of all

kinds are supposed to produce appropriate, sufficiently eloquent language that does
not violate social norms, discriminate against certain groups of people or perpetuate
biases that are incorporated into training data (all of which is especially precarious
in open domain conversations) (Köbis et al., 2021; West et al., 2019; Danaher, 2018;
Silvervarg et al., 2012; Sheng et al., 2019; Bolukbasi et al. 2016).

4.1.2.2  Data Sources  Natural language generation is based on finding statistical pat-
terns in text corpuses (Solaiman et al., 2019), which then allows a machine learning
model, among other things, to predict the next word in a sentence based on previ-
ous words. To learn those patterns, the chosen text corpuses can be digitized books,
forum posts, news articles, communication data, Wikipedia articles, websites, blogs,
scientific papers, and many more.

4.1.2.3  Tracked States and  Traits States and traits that can be tracked in order
to assess text data quality may range from an author’s educational background or
occupation, intelligence, the characteristics of his or her keyboard strokes or display
touching behavior (backspacing etc.), the time between writing and posting, and in
particular by assessing the used publication platform, filtering intermediates, review
processes, and the language skills themselves.

4.1.2.4  Quality Data Contexts  Especially text data that is not produced by profes-
sionals, meaning journalists, writers, scientists, etc., but by lay persons is expected to
be of lower quality. Text data that is not editorially controlled and therefore did not
undergo any kind of review or filtering intermediate may be interspersed with ortho-
graphic mistakes, poor syntax, smaller word pools, slang, invectives, strong biases,
etc. Quality data contexts are to be assessed in dependence on the respective purpose
of an application for natural language generation. Texts from the public domain may
be suited to improve a chatbot’s realism, hence its ability to produce convincing,
authentic, and human-like everyday language. On the other hand, these texts can be
infiltrated with aggressive, discriminatory, or offensive phrases (Wolf et al., 2017).
To avoid these and other pitfalls, the selection of text corpuses that are used to train
conversational robots should not follow the bigger-is-better-approach like many com-
mercially developed chatbots do (Bender et al., 2021). Instead, the selection of cor-
puses can be narrowed down to digital writings that underwent a firm quality check
through publishers, peer reviews, or media agencies, that is embedded in a sophis-
ticated web of citations or links, or that stem from individuals with high levels of
language skills. Moreover, language proficiency can be determined by assessing the

13
576
T. Hagendorff

structure, continuity, errors, vocabulary richness, length of sentence, changes made

to text, etc.

4.1.3 Search Engines

4.1.3.1  Machine Behavior Objectives Modern search engines like Google Search,

Bing, Yandex, etc. use a plethora of signals to rank search results, make autocom-
plete suggestions, predict users’ intentions, evaluate websites, and so on. The main
machine behavior objective is to ensure that rankings and content fit to the anticipated
needs of the users. This, in turn, is supposed to cause a lock-in-effect and bind users to
the respective search engine, eventually raising the likelihood of contact with adver-
tisements. Despite this well-established machine behavior objective, one can name
several other objectives that could determine the architecture of search algorithms
like content quality, expertise, and trustworthiness (in general search engines), equal
opportunity (in people search engines), sustainability (in product search engines),
and many more.

4.1.3.2  Data Sources  Search engines use diverse tracking techniques, harnessing the
large amount of different human–computer-interactions. Each list of search results
shown to users nudges them to become behavioral data contributors for further
calibration and model training by clicking on links, mousing over items, using the
back button, scrolling through pages, entering terms in search bars, interacting with
ads, spending time on a page, and many more. Besides such behavioral data, search
engines can analyze main and supplementary contents of websites, the  amount of
internal and backlinks to a website, labels, page load speed, aggregated views, end-
user device specifications, duplicates, and so on.

4.1.3.3  Tracked States and Traits  Many of today’s relevant search engines are embed-
ded in broader online platforms that allow for a comprehensive user classification. By
collecting and analyzing data on search terms, visited websites, clicked ads, user
location, keyboard strokes, mouse movements, interaction speed, product or profile
views, and the like, it becomes possible to probabilistically infer a variety of differ-
ent personal states and traits. Among them are a person’s gender, age, occupation,
residence, religion, political views, favorite brands, personality, intelligence, literacy,
and many more (Bi et al., 2013). These states and traits can then be used to assess a
user’s “signal quality”.

4.1.3.4  Quality Data Contexts  Professional general search engines do have page
quality rating systems in place (Underwood, 2015). They are used to recognize
the purpose of a website. Beneficial pages that are supposed to help users and are
created by individuals with high expertise, authoritativeness, and trustworthiness
receive the highest ranking. Pages that contain hate or misinformation, encourage
harm to others, or have a deceptive intent receive the lowest rating. However, these
page quality ratings do not solely determine the search results. They are accompa-
nied by machine learning techniques that “digest” user behavior in order to re-train

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 577

the search algorithm. This user behavior can also be assigned to varying “quality”
stages. The clickstream habits of a person who, for instance, regularly uses politi-
cally extremist search terms, visits websites of low quality, has numerous typos,
etc. should be less considered for shaping the search algorithm. On the other hand,
clickstream habits that give evidence of ethically desired traits could preferably be
used to optimize ranking algorithms.

4.1.4 Social Media

4.1.4.1  Machine Behavior Objectives Recommendation systems on social media

platforms come in all shapes and sizes. They are used to filter posts, friends, images,
videos, music, news, search results, and many more. Hitherto, the main goal of these
systems is to increase user engagement in order to bind them to the respective plat-
form. This, in turn, shall raise the likelihood of advertisement contact and click-
through-rates (Eyal & Hoover, 2014; Hagendorff, 2019b; Kuss & Griffiths, 2017).
Taking social responsibility seriously, platforms could rearrange their objectives
towards values of a vital and fair public discourse, truth, and information quality.
This means to change the methods for algorithmic measurement and determination
of information relevance. Fake news, hate speech, extremist content, etc. may cause
the strongest user engagement, but the engagement quantity should not determine the
subsequent dissemination and recommendation of the respective content. Instead,
engagement quality should determine data quality and help to build responsible
machine recommendation behavior.

4.1.4.2  Data Sources  Social media platforms can track a plethora of user signals.
Amongst the more obvious ones are clickstreams, search queries, demographic
or profile information, reactions to posts, duration of post views, scroll behavior,
networks of friends, comments, and many more. All these data traces are used to
determine the relevance of posts, videos, images, tweets, friend suggestions, etc.
in order to operate the platforms’ recommendation systems.

4.1.4.3  Tracked States and Traits  Tracing back to a differentiation from behavioral

economics (Kahneman, 2012), one can distinguish system-1- and system-2-inter-
actions. System-1 comprises fast, emotional, effortless, cognitively simple think-
ing processes that are prone to biases and mistakes, whereas system-2 covers slow,
rational, and deliberate thinking processes. Those two modes of thinking do also
influence the way digital platforms are used (Lischka & Stöcker, 2017). Amongst
other factors, social media platforms could measure whether users operate with a
platform on a more irrational, bias prone, impulsive system-1 mode. This mode
allows for rather quick and impulsive actions, resulting in a stream of unreflected
human–computer-interactions. Impulsive, system-1 user behavior could be tracked
by things like reaction or comment speed, the susceptibility to nudging techniques,
and scrolling or reading behavior. The platforms could also use further inferences
to educational levels, intelligence, psychological traits and states like anxieties or

13
578
T. Hagendorff

radical political or religious views to assess user behavior that is connected to the
respective attributes.

4.1.4.4  Quality Data Contexts  Cognitive heuristics that are part of system-1-in-
teractions influence the way individuals interact with social media. Hence, biases
are technically perpetuated via recommendation systems (Stieglitz & Dang-Xuan,
2012, 2013). Quality data can thus be scraped from contexts where user gener-
ated data do mainly represent system-2-human–computer-interactions. This way,
recommendation systems can be trained on behavioral data that represents fewer
biases and impulsive reactions. Instead of negatively affecting public discourse by
helping the spreading of content that is mostly suited to cause emotional arousal
and impulsive reactions, platforms help to automatically disseminate content that
is less “toxic” for public discourse. In this context, a special focus can be laid
on so-called “superusers”, who are not just very active users with high levels of
engagement, but who also disproportionately spread misinformation. According to
one rare source, internal committees at Facebook urged to lower recommendation
scores for content posted by “superusers” on the far right or far left of the politi-
cal spectrum. Content from moderate users, in turn, would receive higher scores
(Ng, 2020). This advice was turned down by Facebook’s leadership. However, it
perfectly corresponds to the idea that is advocated here. When recommendation
systems are trained on behavioral data from individuals with higher education,
who do not represent political or religious extremes, and who normally interact
with quality, i.e. journalistically or scientifically verified, trustworthy information,
it is to be expected that those systems automatically spread content that comes
with various benefits for the public, instead of harms, leading to a situation where
everyone is better off.

4.1.5 Online Shopping

4.1.5.1  Machine Behavior Objectives E-commerce platforms where people can

buy goods and services use various methods to promote purchasing behavior. They
use shopping search engines, product recommendation systems, product reviews,
dynamic pricing, cross selling, customer analytics tools, conversion rate optimiza-
tion, conversion funnels, varying payment options, specific user interface designs,
etc. in order to maximize revenues. Therein lies the entrenched main machine behav-
ior objective. However, via tweaking the underlying machine learning algorithms,
the machine behavior objective can be diversified, comprising not just the pursuit of
economic values, but also values of sustainability or public health. Especially the big-
gest online stores like Amazon, eBay, Walmart, Jingdong, Alibaba, and others would
cause a significant impact by just slightly changing their machine learning models
towards the mentioned values, taking their corporate social responsibility seriously.

4.1.5.2  Data Sources  No different than search engines or social media platforms,
online retailers can collect a broad variety of user data. They can analyze and track

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 579

the number of transactions, all kinds of product and customer data, the conversion
rate, product impressions, average order values, product detail views, adding or
removing of products from the shopping basket, withdrawals from checkout pro-
cess, customer lifetime values, traffic sources, details of users’ devices, and many
more.

4.1.5.3  Tracked States and Traits Using the many data sources e-commerce plat-
forms can gather as a basis, they can implement specific automated mechanisms for
customer segmentation. Typically, differentiations for types of customers are made
purely from a sales perspective, distinguishing between loyal, impulsive, novice, etc.
customers. Notwithstanding that, customers can be segmented along criteria like
health- or eco-consciousness by analyzing their product views, shopping behavior,
product reviews, search terms, personality, socio-demographic factors, and the like.

4.1.5.4  Quality Data Contexts Sticking to the aforementioned values of sustain-

ability and public health, e-commerce platforms could use all available data from
health- as well as eco-conscious customers and use specifically those data to train
models for product recommendations, dynamic pricing, or ranking algorithms for
their search engines, to name just three major setting options. Such measures could
significantly foster the extent to which e-commerce platforms promote more sustain-
able and healthier consumer behavior.

5 Discussion

In the following section, I want to gather some major points for discussion to the
suggested approach to achieve beneficial machine learning applications. The
approach relies on several practices like tracking, profiling, ranking, or filtering
that have been applied in contexts of technology misuse for illegitimate ends more
than just a few times (Brundage et al. 2018; Crawford et al. 2019; Crawford, 2021;
Engelmann et al., 2019). Nevertheless, the mentioned methods cannot be simply dis-
carded. Rather, they can be used for purposes that are in line with the common good
and ethical values. The following subchapters will elaborate on that and take up the
seven most pressing concerns that can be raised when putting the paper’s ideas into
practice.

5.1 Ranking Behavior

First of all, I want to address some ethical concerns that are connected with the idea
of ranking human behavior, as described in chapter 3. As a cautionary tale how such
ranking practices fail one could point at digital marketing, where marketers consider
customers to be either “targets” or “waste” (Turow, 2012). People’s tastes, demo-
graphic profiles, beliefs, income levels, and many more are digitally measured for
the purpose of discriminating them for commercial goals and to find the customers

13
580
T. Hagendorff

that are deemed most valuable. Reputation silos are constructed around people who
statistically seem similar, but this practice often circumvents ethoses around equal
opportunity, justice, or transparency. However, the idea of ranking or sorting human
behavior along certain dimensions or hierarchies is not to be refused altogether. One
can put the concept in another light when reframing it according to assessment cri-
teria from ethics. Hence, besides arguing for a repurposing of existing social sort-
ing structures in marketing for other socially accepted ends, one can in general say
that society embraces various practices of behavior assessment under the term “eth-
ics”. Here, one transitions from saying that some social positions, dispositions, or
stages of development are better than others to classifying behavior as morally right
or wrong. In the context of this paper, though, the argument is somewhat more spe-
cific: Currently, digital tracking systems that measure users’ social as well as psy-
chological traits and states are used mainly to support marketing decisions, to foster
customer relationship management, to personalize the marketing mix to individuals,
and to support many other commercial purposes (Wedel & Kannan, 2016). I argue
that the rich toolset that is already established for marketing purposes can be repur-
posed to assess the ethical quality of data contexts in order to develop beneficial
machine learning applications. Currently, tracking methods are not deployed to
evaluate digital behavior from an ethical point of view, which would be essential
to assess ethical data quality dimensions. With that said, this assessment does not
necessarily require in-depth knowledge about ethical theories that are developed in
philosophic discourses. According to the intuitionist approach in moral psychology,
ethical judgements are driven primarily by one’s intuitions (Haidt, 2001). In view of
that and apart from complicated, dilemmatic cases, moral intuitions can appropri-
ately guide ethical reasoning. This becomes evident when considering the tacit con-
sensus about ethical target values or “attraction poles” that are embedded into socio-
logical as well as psychological theories. Normative assumptions about the value
of prosocial dispositions, rationality, moral development, openness, impulse control,
positive emotionality, and the like are seldomly contested. Hence, these intuitive
normative assumptions can in many cases guide data quality evaluations. This shall
not downplay the potential complexities and conflicts that can be connected to such
an evaluation or to AI technology assessments per se (Raji et  al. 2020; Suresh &
Guttag, 2020). But as a first step, moral intuitions might be a good “compass” to
guide decisions on which features in data sets should be weighted stronger than oth-
ers in order to promote beneficial model training.

5.2 Paternalism

Another potential objection to the ideas presented here is that they depict a form of
ethical paternalism. The decision about what defines quality data contexts is made
by machine learning practitioners and other members off the respective coalition of
stakeholders that become involved in deciding which values and normative assump-
tions are to be implemented into technology. The decisions made in this coalition
affect other collectives without their democratic consent. The idea of paternalis-
tic thinking is that some individuals are more competent, rational, or versed than

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 581

others and that the former can decide for the latter for their advantage. Paternal-
istic approaches are mainly criticized because they not only limit the freedom of
affected individuals, but this is also done without their consent (Sartorius, 1983). In
this context, two main counterarguments can be raised. First, the idea of developing
beneficial supervised machine learning applications follows, as the term suggests,
the notion of being beneficial or advantageous for as many individuals as possible,
hence promoting the common good. Second, machine learning applications do in
nearly all cases affect individuals without their explicit consent or knowledge. Val-
ues are part of technologies itself (Brey, 2010), and the process of embedding cer-
tain values or choosing architectural designs is in many regards not a democratic
one. Rather, a small group of technology developers possess the power to make far
reaching decisions for a group of end users that can comprise millions or billions
of individuals (Lessig, 2006). The decisive question is whether values are embed-
ded in software in an arbitrary and nonreflective way, often perpetuating prejudice,
biases, or misunderstandings, or whether these values and ethical norms are chosen
carefully and consciously. Here, I opt for the latter. In addition to that, technology
ethics indeed typically operates with paternalistic, top-down norms and standards.
Asimov’s Three Laws for Robots, for instance, is presumably the most well-known
approach (Asimov, 2004). However, in the context of this paper, the paternalistic
top-down approach is not solely embraced but combined with one that is bottom-
up. Bottom-up approaches mean that a technical agent, in this case machine learn-
ing models, explores courses of action that represent morally praiseworthy exam-
ples (Wallach & Allen, 2009). Hence, the agent achieves “moral capabilities” by
surveying its environment, similar to childhood development. Here, both top-down
and bottom-up approaches are integrated. A top-down analysis is done with regard
to the way training data is filtered or selected according to ethical criteria, while on
the other hand a developmental or bottom-up approach is chosen in order to allow
the machine to learn a certain behavior from data sets.

5.3 Transparency

Another objection, that is akin to the one above, is to stress that the proposed
approach for beneficial machine learning applications is non-transparent, leaving
affected individuals unwitting about the technical measures that are conditioning
their user experience and filtering mechanisms for quality data. The counterargu-
ment to this objection is that platforms or software developers could and should have
no problems whatsoever making transparency statements, thereby informing poten-
tial reviewers about the value and design choices, ultimately making them subject to
public scrutiny. This would show that definitions about ethical quality dimensions
of training data are in line with cultural consensuses, ethical theories, and moral
intuitions. Revisiting the above-mentioned examples from chapter  4.1, the major-
ity of people would consent to the claim that autonomous cars should be safe, that
natural language generation should avoid simplicity and perform eloquently, that
e-commerce platforms should promote sustainable shopping routines, that recom-
mendation systems on social media platforms as well as search engines should not

13
582
T. Hagendorff

foster political extremism and information that is “toxic” to the public discourse but
promote quality content, expertise, and truth. In general, it should be a necessary
prerequisite that through transparency statements the criteria by which data quality
is evaluated are described and justified when putting technologies for digital behav-
ior tracking as well as machine learning models that are trained with behavioral data
in place.

5.4 Privacy

In order to select for quality data or data that represents ethical behavior, various
tracking or surveillance techniques must be in place. This leads to a further objec-
tion, namely that the proposed approach for beneficial machine learning application
goes hand in hand with privacy violations. This criticism requires a more detailed
examination. First of all, although this might be a weak argument, the proposed
approach does not necessarily opt for an extension of methods for recording behav-
ioral data that are already entrenched (D’Mello & Kory, 2015; Zeng et al., 2009).
Here, it is important that these methods are used for legitimate ends, not that they
are abolished (Belliger & Krieger, 2018; Hagendorff, 2019a). As already discussed
in a previous chapter, the idea of classifying people or people’s behavior raises
weighty ethical questions and has its ailments, especially with regard to the fea-
sibility of data-driven assessments of sensitive traits like mental illnesses, intelli-
gence, personality, and the like. However, data protecting measures that would pro-
hibit these assessments like the one mentioned are primarily aiming at preventing
unjust discrimination and at securing personal autonomy (Roßnagel, 2007). In the
end, though, is it important to remember that from the pure existence of these tech-
nologies alone it does not necessarily follow that they are misused. On the contrary,
when binding legal norms as well as strong ethical tenets are entrenched, tracking
and profiling can be used for the common good, as it is described here.
Moreover, techniques for tracking user behavior and assessing behavioral data
quality can work by only using anonymized and aggregated data, avoiding oppor-
tunities to identify certain individuals. Although current trends in informational pri-
vacy research are extending the idea of privacy violations from a merely individu-
alistic data protection perspective (Westin, 1967) towards notions of interdependent,
group, collective, or predictive privacy (Biczók & Chia, 2013; Mühlhoff, 2021; Yu
& Grossklags, 2016), the main “type” of privacy violation still remains in the form
of a direct access or probabilistic inference to sensitive information that are tied to
a particular person. But this “classical” form of privacy violation can be avoided by
anonymization techniques, mainly differential privacy or k-anonymization (Dwork
et al. 2006; Dwork 2008; Dwork & Roth, 2013; Samarati & Sweeney, 1998). These
techniques imply to alter datasets by adding noise or by manipulating them in a way
that individuals theoretically cannot be identified. That does not necessarily provide
full privacy or full protection against re-identification, though. Moreover, making
datasets more privacy preserving comes at a price, namely accuracy. On the flipside,
implementing differential privacy can lead to access to new data sources that hith-
erto could not be collected due to privacy concerns (Kearns & Roth, 2020, p. 56).

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 583

In this way, privacy enhancing measurements can actually lead to an increase in the
amount of sensitive behavioral data, which can then be used to assess training data
quality from an ethical perspective. Eventually, the proposed approach, which is
indeed very data-intensive, can go hand in hand with individual privacy preserving
measurements.

5.5 Accuracy

Another objection is that techniques to digitally assess and rate human behavior may
be inaccurate and create false positives and negatives. These techniques as well as
data traces per se construct rather than represent an individual’s true actions, traits,
and states. Accordingly, behavioral data as well as the computational processing
thereof cannot be condensed in information that represents “reality”. Rather, differ-
ent “realities” can be constructed from data and algorithms (Lewis, 2015; Matzner,
2016). They do not work impassive, but shape how we understand the world in a
performative manner (Kitchin, 2017), while allowing probabilistic inferences on
in situ behavior. In individual cases, this can lead to detrimental false positives or
negatives. But the methods proposed in this paper all operate with aggregated and,
in the best case, anonymized data, which means that individuals face unjust techno-
logical consequences only when tracking and profiling techniques fail significantly.
Only in the unlikely event they come up with misclassifications in an overwhelming
number of cases, the selection of quality training data and therefore the trained mod-
els would become skewed.

5.6 Discrimination

Akin to the aforementioned objection is the argument against algorithmic discrimi-

nation. The ideas presented in this paper could fall prey to such arguments since spe-
cific biases in data sets are intendedly promoted, resulting in “skewed” algorithmic
decision making. Previous discourses on algorithmic discrimination rightly criticize
that machine learning techniques perpetuate existing biases that are entrenched in
data sets and therefore foster unfair discrimination. Under the umbrella term “fair-
ness, accountability, and transparency in machine learning” (FAT ML), machine
learning practitioners collect methods for reducing algorithmic discrimination pri-
marily by dealing with protected attributes like gender, age, ethnicity, etc. (Dwork
et  al. 2011; Kleinberg et  al., 2016; Veale & Binns, 2017). However, here I argue
that one should reintroduce or promote “algorithmic discrimination”, but, needless
to say, not in the traditional way. Data sets may contain features that are critical in a
way that they should be weighted stronger than others. This can be the case even if
these features perpetuate biases, since biases can actually be necessary for fairness.
Dutta et al. (2020) give the example of a hiring for fire fighters, where candidates
should be able to lift heavy weights, which leads to a preference for men rather than
women. In short, biases are acceptable if they are critical for the legitimate solution
of a given task. Here, I propose to promote biases in data sets used to train machine
learning models that lead to a preferability of features that are desirable from an

13
584
T. Hagendorff

ethical point of view. At the same time, however, this means to put individuals at a
“disadvantage” who produce behavioral data that originates in deeds, personal traits,
or mental states that are socially less esteemed like risky behavior, detrimental norm
violations, bad language, low education, political or religious extremism, flawed log-
ical reasoning, impulsiveness, and the like. This way, machine behavior that results
from recognizing statistical patterns in behavioral data is not “socialized” by general
populations (n = all) but by specific subgroups (n = X, where X ⊊ all) that comprise
individuals who are most competent, eligible, or morally versed for a particular task.
While the typical notion of algorithmic discrimination is pointing towards unfair
computational outputs, the kind of algorithmic discrimination that is proposed here
aims at introducing stricter filters that thwart particular data traces to become train-
ing data for machine learning. This way computational outputs manifest values that
correspond to ethical virtues and that are socially accepted, appreciated, and sought-
after like friendliness, literacy, truthfulness, positive emotionality, prosocial orien-
tations, etc. Selecting for these values is a task that comes with a heavy responsi-
bility. That is why it should not only be put on the shoulders of machine learning
practitioners. Such a task can best be addressed when building a coalition of stake-
holders, when combining technical research with the social sciences, psychology,
domain-specific applied ethics, standard setting bodies, etc. As a rule of thumb, the
extend and inclusiveness of this coalition correlates with the level of criticality of
a particular application of algorithmic decision making (Wendehorst et  al., 2019).
By following such an integrated approach, unintended cognitive biases in the selec-
tion off intended training data biases can be tracked, but it can also be ensured that
the selection of values and traits stands in accordance with democratic consensuses
(Floridi et al., 2018; Brundage et al. 2020; Daly et al. 2019b; Danaher et al. 2017;
Spindler et al., 2020).

5.7 Systemic Imperatives

A further objection is to remark that beneficial machine learning applications, as

supposed in this paper, stand in contradiction to systemic imperatives and goals of
the economy. While making autonomous cars safer should result in having a com-
petitive advantage, rendering recommendation systems on e-commerce platforms
towards promoting sustainable, but more expensive products or on social media
platforms towards less engaging content means that the platforms acquire fewer
purchasers or users who can be influenced by online advertisements. However, the
point of making something beneficial, in this case machine learning applications,
is to overwrite systemic imperatives in case they have detrimental effects for par-
ticular individuals or society at large. Being successful according to the logic of a
certain social system (Luhmann, 1995) does not necessarily mean that this success
is morally justified (Habermas, 1987). This holds especially true with regard to the
economy (Brand & Wissen, 2017). On a related note, beneficial machine learning
applications, which are traditionally used in areas like health or crisis response,
satellite image interpretation, climate action, poverty reduction, wildlife preserva-
tion, and the like (Chui et al., 2018) do in many cases not follow systemic but moral

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 585

imperatives. What is special in the case of the ideas presented here is that I do not
propose to invent new machine learning application in hitherto undiscovered fields
of society. Rather, I opt for reshaping applications that are already entrenched in
areas that are structured by systemic imperatives and eventually aim at being prof-
itable. In contexts that are purely dominated by monetary considerations, it is of
course difficult to put the ideas into practice. Nevertheless, companies should at
no point see the pursuit of profits as their only target. And as soon as they include
social responsibility into their repertory of values, they can embrace the presented
approach for beneficial machine learning.

6 Conclusion

In their classic book “Moral Machines”, Wallach and Allen state: “The vision of
learning systems developing naturally toward an ethical sensibility that values
humans and human ethical concerns is an optimistic vision […].” (Wallach & Allen,
2009, p.  110) This paper is a tangible proposal how this vision could be put into
practice. It stresses the importance of “feeding” machine learning applications not
with all relevant behavioral data that is available, but with a particular selection
of it, namely with quality data. Following the typical big data approach and using
all available data to train models can have detrimental effects. This can not only
be shown by, for instance, pointing at various cases of algorithmic discrimination.
It only recently got obvious when COVID-19 caused dramatic changes in online
shopping and other digitally recorded behavior, so that its inclusion in training sets
caused machine learning applications to malfunction, making manual interventions
necessary (Heaven, 2020). Thus, more rigorous mechanisms to filter training data
sets have to be put in place, ensuring that, among others, only “good data” become
training stimuli, meaning that digitally recorded behavior is classified and assessed
along ethical criteria. Moral machine behavior is dependent on moral human behav-
ior. Hence, both have to be linked. Incidentally, this idea is in accordance with the
proposal to found “computational psychiatry” where computational problems like
malfunctions or machine biases are seen in analogy to mental disorders (Schulz &
Dayan, 2020). Taking up this proposal, this paper can be seen as an initial sugges-
tion on how to give “therapy” to machine learning models.
Many machine learning applications acquire their “intelligence” by “captur-
ing” (Mühlhoff, 2019, 1873) aggregated human cognitive and behavioral abilities.
Hitherto, these aggregations of recordings of human behavior are hardly presorted
before becoming training stimuli for machine behavior. This paper is a plea to do so
and thereby to achieve truly beneficial machine learning. Its arguments start at the
assumption that a person’s social background, dispositions, educational level, etc.
shapes his or her way of using digital devices. In turn, those devices are able to
track, infer, and measure a user’s personal states or traits. Depending on these attrib-
utes, an ethical assessment of data traces the respective user produces, namely the
data traces that provide the basis to the measurements itself as well as data traces
that are situated in the same context as the measurement, can take place. Subse-
quently, this assessment enables to single out quality data that are representing

13
586
T. Hagendorff

behavior of individuals who are deemed to be especially competent, eligible, or

morally versed for a particular task. This method for sampling out particular train-
ing stimuli stands in contrast to the n = all ideology. It serves to tailor training data
in a way that machine behavior can correspond to values of the common good and
become truly beneficial.

Acknowledgements  This research was supported by the Cluster of Excellence “Machine Learning – New
Perspectives for Science” funded by the Deutsche Forschungsgemeinschaft (DFG, German Research
Foundation) under Germany’s Excellence Strategy—Reference Number EXC 2064/1—Project ID
390727645. I would like to thank Sarah Fabi, Zeynep Akata, Ulrike von Luxburg, and Sebastian Bordt for
very helpful comments on the manuscript.

Funding  Open Access funding enabled and organized by Projekt DEAL.

Open Access  This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as
you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-
mons licence, and indicate if changes were made. The images or other third party material in this article
are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is
not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission
directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​
ses/​by/4.​0/.

References
Abdoli, N., Farnia, V., Delavar, A., Esmaeili, A., Dortaj, F., Farrokhi, N., et al. (2015). Poor mental health
status and aggression are associated with poor driving behavior among male traffic offenders. In
Neuropsychiatric Disease and Treatment, 11, 2071–2078.
Acar, Gunes; Eubank, Christian; Englehardt, Steven; Juarez, Marc; Narayanan, Arvind; Diaz, Claudia
(2014): The Web Never Forgets. Persistent Tracking Mechanisms in the Wild. In Gail-Joon Ahn,
Moti Yung, Ninghui Li (Eds.): Proceedings of the 2014 ACM SIGSAC Conference on Computer
and Communications Security - CCS ’14. ACM SIGSAC Conference. Scottsdale, Arizona. New
York: ACM Press, pp. 674–689.
Amodei, Dario; Olah, Chris; Steinhardt, Jacob; Christiano, Paul; Schulman, John; Mané, Dan (2017):
Concrete Problems in AI Safety. In arXiv, pp. 1–29.
Amoore, Louise (2011): Data Derivatives. On the Emergence of a Security Risk Calculus for Our Times.
In Theory, Culture & Society 28 (6), pp. 24–43.
Anstey, K. J., Wood, J., Lord, S., & Walker, J. G. (2005). Cognitive, sensory and physical factors enabling
driving safety in older adults. In Clinical Psychology Review, 25(1), 45–65.
Asimov, I. (2004). I, Robot. Random House LLC.
Barocas, S., & Selbst, A. D. (2016). Big data’s disparate impact. In California Law Review, 104, 671–732.
Belliger, Andréa; Krieger, David J. (2018): Network Public Governance. On Privacy and the Informa-
tional Self. Bielefeld: Transcript.
Bender, Emily M.; Gebru, Timnit; McMillan-Major, Angelina; Mitchell, Margaret (2021): On the Dan-
gers of Stochastic Parrots: Can Language Models Be Too Big?: ACM, pp. 1–14.
Bi, Bin; Shokouhi, Milad; Kosinski, Michal; Graepel, Thore (2013): Inferring the demographics of search
users. In Daniel Schwabe, Virgílio Almeida, Hartmut Glaser, Ricardo Baeza-Yates, Sue Moon
(Eds.): Proceedings of the 22nd International Conference on World Wide Web - WWW ’13. New
York: ACM Press, pp. 131–140.
Biczók, G., & Chia, P. H. (2013). Interdependent privacy: Let me share your data. Springer.
Bogard, William (1996): The Simulation of Surveillance. Hypercontrol in Telematic Societies. Cam-
bridge: Cambridge University Press.

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 587

Bourdieu, Pierre (1984): Distinction. A Social Critique of the Judgement of Taste. Cambridge, Massa-
chusetts: Harvard University Press.
Bourdieu, P. (1989). Social space and symbolic power. In Sociological Theory, 7(1), 14–25.
boyd, danah, . (2012). White Flight in Networked Publics. How Race and Class Shaped American Teen
Engagement with MySpace and Facebook. In L. Nakamura & P. A. Chow-White (Eds.), Race After
the Internet (pp. 203–222). Routledge.
Bolukbasi, Tolga; Chang, Kai-Wei; Zou, James; Saligrama, Venkatesh; Kalai, Adam (2016): Man is
to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings. In arXiv,
pp. 1–25.
Brand, Ulrich; Wissen, Markus (2017): Imperiale Lebensweise. Zur Ausbeutung von Mensch und Natur
im globalen Kapitalismus. München: oekom Verlag.
Brey, P. (2010). Values in technology and disclosive computer ethics. In L. Floridi (Ed.), The Cambridge
Handbook of Information and Computer Ethics (pp. 41–58). Cambridge University Press.
Calvo, R. A., Peters, D., & Cave, S. (2020). Advancing impact assessment for intelligent systems. In
Nature Machine Intelligence, 1, 1–3.
Chui, Michael; Harryson, Martin; Manyika, James; Roberts, Roger; Chung, Rita; van Heteren, Ashley;
Nel, Pieter (2018): Notes from the AI Frontier. Applying AI  for Social Good. McKinsey Global
Institute: McKinsey&Company, pp. 1–52.
Clarke, R. (1988). Information technology and dataveillance. In Communications of the ACM, 31(5),
498–512.
Clarke, R. (1994). The digital persona and its application to data surveillance. In the Information Society,
10(2), 77–92.
Crawford, Kate (2021): Atlas of AI. Power, Politics, and the Planetary Costs of Artificial Intelligence.
New Haven: Yale University Press.
Crawford, Kate; Dobbe, Roel; Dryer, Theodora; Fried, Genevieve; Green, Ben; Kaziunas, Elizabeth et al.
(2019): AI Now 2019 Report. AI Now. New York. Available online at https://​ainow​insti​tute.​org/​
AI_​Now_​2019_​Report.​pdf, checked on 12/18/2019.
Čubranić-Dobrodolac, M., Lipovac, K., Čičević, S., & Antić, B. (2017). A model for traffic accidents pre-
diction based on driver personality traits assessment. In PROMET, 29(6), 631–642.
Daly, Angela; Hagendorff, Thilo; Hui, Li; Mann, Monique; Marda, Vidushi; Wagner, Ben et al. (2019b):
Artificial Intelligence, Governance and Ethics:  Global Perspectives. The Chinese University of
Hong Kong Faculty of Law Research Paper No. 2019–15. In SSRN Journal, pp. 1–41.
Daly, A., Devitt, K. S., & Mann, M. (Eds.). (2019a). Good Data. Institute of Network Cultures.
Danaher, J. (2018). Toward an ethics of AI assistants. An Initial Framework. in Philos. Technol., 31(4),
629–653.
Danaher, John; Hogan, Michael J.; Noone, Chris; Kennedy, Rónán; Behan, Anthony; Paor, Aisling de
et al. (2017): Algorithmic governance. Developing a research agenda through the power of collec-
tive intelligence. In Big Data & Society 4 (2), 205395171772655.
Davidow, William; Malone, Michael S. (2020): Don’t Regulate Artificial Intelligence: Starve It (Scien-
tific American). Available online at https://​blogs.​scien​tific​ameri​can.​com/​obser​vatio​ns/​dont-​regul​
ate-​artif​i cial-​intel​ligen​ce-​starve-​it/, checked on 5/8/2020.
Deleuze, Gilles (1992): Postscript on the Societies of Control. In October 59, pp. 3–7.
D’Mello, S. (2013). A selective meta-analysis on the relative incidence of discrete affective states during
learning with technology. In Journal of Educational Psychology, 105(4), 1082–1099.
D’Mello, S., & Kory, J. (2015). A review and meta-analysis of multimodal affect detection systems. In
ACM Comput. Surv., 47(3), 1–36.
Domingos, Pedro (2015): The Master Algorithm. How the Quest for the Ultimate Learning Machine Will
Remake Our World. New York: Basic Books.
Dutta, Sanghamitra; Venkatesh, Praveen; Mardziel, Piotr; Datta, Anupam; Grover, Pulkit (2020): An
Information-Theoretic Quantification of Discrimination with Exempt Features. AAAI Conference
on Artificial Intelligence, pp. 1–28.
Dwork, Cynthia (2008): Differential Privacy: A Survey of Results. In Manindra Agrawal, Dingzhu Du,
Zhenhua Duan, Angsheng Li (Eds.): Theory and Applications of Models of Computation. Berlin:
Springer Berlin Heidelberg, pp. 1–19.
Dwork, C., et al. (2006). Differential Privacy. In D. Hutchison, T. Kanade, J. Kittler, J. M. Kleinberg, F.
Mattern, & J. C. Mitchell (Eds.), Automata, Languages and Programming (pp. 1–12). Springer.
Dwork, C., & Roth, A. (2013). The algorithmic foundations of differential privacy. In FNT in Theoretical
Computer Science, 9(3–4), 211–407.

13
588
T. Hagendorff

Dwork, Cynthia; Hardt, Moritz; Pitassi, Toniann; Reingold, Omer; Zemel, Richard (2011): Fairness
Through Awareness. In arXiv, pp. 1–24.
Eady, T. A. (2019): Why Tesla’s Fleet Miles Matter for Autonomous Driving (Medium). Available online
at https://​towar​dsdat​ascie​nce.​com/​why-​teslas-​fleet-​miles-​matter-​for-​auton​omous-​drivi​ng-​8e485​
03a46​2f, checked on 5/11/2020.
Engelmann, Severin; Chen, Mo; Fischer, Felix; Kao, Ching-yu; Grossklags, Jens (2019): Clear Sanctions,
Vague Rewards: How China’s Social Credit System Currently Defines "Good" and "Bad" Behav-
ior. In Proceedings of the Conference on Fairness, Accountability, and Transparency - FAT* ’19,
pp. 69–78.
Erikson, E. H. (1980). Identity and the life cycle. W.W. Norton.
Erikson, R., Goldthorpe, J. H., & Portocarero, L. (1979). Intergenerational class mobility in three West-
ern European societies: England, France and Sweden. In the British Journal of Sociology, 30(4),
415–441.
Eubanks, Virginia (2018): Automating Inequality. How High-Tech Tools Profile, Police, and Punish the
Poor. New York: St. Marting’s Press.
Eyal, Nir; Hoover, Ryan (2014): Hooked. How to build Habit-Forming Products. Princeton, NJ: Prince-
ton University Press.
Fergusson, D., Swain-Campbell, N., & Horwood, J. (2003). Risky driving behaviour in young people:
prevalence, personal characteristics and traffic accidents. In Australian and New Zealand Journal
of Public Health, 27(3), 337–342.
Floridi, Luciano; Cowls, Josh; Beltrametti, Monica; Chatila, Raja; Chazerand, Patrice; Dignum, Virginia
et  al. (2018): AI4People - An Ethical Framework for a Good AI Society. Opportunities, Risks,
Principles, and Recommendations. In Minds and Machines 28 (4), pp. 689–707.
Friedman, B., & Nissenbaum, H. (1996). Bias in computer systems. ACM Transactions on Information
Systems, 14(3), 330–347.
Gebru, Timnit; Morgenstern, Jamie; Vecchione, Briana; Vaughan, Jennifer Wortman; Wallach, Hanna;
Daumeé, Hal, III; Crawford, Kate (2018): Datasheets for Datasets. In arXiv, pp. 1–17.
Ghani, Rayid; Schierholz, Malte (2017): Machine Learning. In Ian Foster, Rayid Ghani, Ron S. Jarmin,
Frauke Kreuter, Julia Lane (Eds.): Big Data and Social Science. A Practical Guide to Methods and
Tools. Boca Raton: CRC Press, pp. 147–186.
Gitelman, L. (Ed.). (2013). “Raw Data” Is an Oxymoron. The MIT Press.
Grusky, David B. (2019): Social stratification. Class, race, and gender in sociological perspective. Lon-
don: Routledge.
Gudivada, V., Apon, A., & Ding, J. (2017). Data quality considerations for big data and machine learn-
ing: going beyond data cleaning and transformations. In International Journal on Advances in Soft-
ware, 10, 1–20.
Guntuku, S., & Chandra; Yaden, David B., Kern, Margaret L., Ungar, Lyle H., Eichstaedt, Johannes C. .
(2017). Detecting depression and mental illness on social media: an integrative review. In Current
Opinion in Behavioral Sciences, 18, 43–49.
Habermas, Jürgen (1987): Theorie des kommunikativen Handelns. 2 volumes. Frankfurt a.M: Suhrkamp
(1).
Hagendorff, Thilo (2019c): Maschinelles Lernen und Diskriminierung. Probleme und Lösungsansätze. In
Österreichische Zeitschrift für Soziologie 44 (1), pp. 53–66.
Hagendorff, T. (2019a). From privacy to anti-discrimination in times of machine learning. In Ethics and
Information Technology, 33(3), 331–343.
Hagendorff, T. (2019b). Jenseits der puren Datenökonomie - Social-Media-Plattformen besser designen.
In C. Ochs, M. Friedewald, T. Hess, & J. Lamla (Eds.), Die Zukunft der Datenökonomie (pp. 327–
342). Springer.
Haggerty, K. D., & Ericson, R. V. (2000). The surveillant assemblage. In the British Journal of Sociol-
ogy, 51(4), 605–622.
Haidt, Jonathan (2001): The Emotional Dog and Its Rational Tail. A Social Intuitionist Approach to
Moral Judgment. In Psychology Review 108 (4), pp. 814–834.
Harari, G. M., Müller, S. R., Aung, M. S. H., & Rentfrow, P. J. (2017). Smartphone sensing methods for
studying behavior in everyday life. In Current Opinion in Behavioral Sciences, 18, 83–90.
Hargittai, E. (2008). The Digital Reproduction of Inequality. In D. B. Grusky (Ed.), Social Stratification
(pp. 936–944). Westview Press.

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 589

Hart, D., Hofmann, V., Edelstein, W., & Keller, M. (1997). The relation of childhood personality types to
adolescent behavior and development: a longitudinal study of Icelandic children. In Developmental
Psychology, 33(2), 195–205.
Heaven, Douglas Will (2020): Our weird behavior during the pandemic is messing with AI models (MIT
Technology Review). Available online at https://​www.​techn​ology​review.​com/​2020/​05/​11/​10015​63/​
covid-​pande​mic-​broken-​ai-​machi​ne-​learn​ing-​amazon-​retail-​fraud-​humans-​in-​the-​loop/, checked
on 6/15/2020.
Hu, Jian; Zeng, Hua-Jun; Li, Hua; Niu, Cheng; Chen, Zheng (2007): Demographic prediction based on
user’s browsing behavior. In Carey Williamson, Mary Ellen Zurko, Peter Patel-Schneider, Prashant
Shenoy (Eds.): Proceedings of the 16th International Conference on World Wide Web - WWW
’07. New York: ACM Press, pp. 151–160.
Hyman, M. M. (1968). Accident vulnerability and blood alcohol concentrations of drivers by demo-
graphic characteristics. In q. J. Stud. Alcohol Suppl., 29(S4), 34–57.
Irani, L. (2016). The hidden faces of automation. In XRDS, 23(2), 34–37.
Irvin, Jeremy; Rajpurkar, Pranav; Ko, Michael; Yu, Yifan; Ciurea-Ilcus, Silviana; Chute, Chris et  al.
(2019): CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Com-
parison. In arXiv, pp. 1–9.
Jo, Eun Seo; Gebru, Timnit (2019): Lessons from Archives. Strategies for Collecting Sociocultural Data
in Machine Learning. In arXiv, pp. 1–11.
John, Oliver P.; Naumann, Laura P.; Soto, Christopher J. (2008): Paradigm Shift to the Integrative Big
Five Trait Taxonomy. History, Measurement, and Conceptual Issues. In Oliver P. John, Richard W.
Robins, Lawrence A. Pervin (Eds.): Handbook of Personality. Theory and Research. New York:
The Guilford Press, pp. 114–158.
Jordan, Michael I.; Mitchell, Tom M. (2015): Machine learning. Trends, perspectives, and prospects. In
Science 349 (6245), pp. 255–260.
Kahneman, D. (2012). Thinking, fast and slow. Penguin.
Kant, Immanuel (1977): Kants Werke, Akademie Textausgabe. Anmerkungen der Bände I-[IX]: Walter
de Gruyter.
Kearns, Michael; Roth, Aaron (2020): The Ethical Algorithm. The Science of Socially Aware Algorithm
Design. New York: Oxford University Press.
Khan, Iftikhar Ahmed; Brinkman, Willem-Paul; Fine, Nick; Hierons, Robert M. (2008): Measuring per-
sonality from keyboard and mouse use. In Joaquim Jorge (Ed.): Proceedings of the 15th European
Conference on Cognitive Ergonomics the Ergonomics of Cool Interaction - ECCE ’08. New York:
ACM Press, pp. 1–8.
Kitchin, Rob; Dodge, Martin (2011): Code/Space. Software and Everyday Life. Cambridge, Massachu-
setts: The MIT Press.
Kitchin, R. (2017). Thinking critically about and researching algorithms. In Information, Communication
& Society, 20(1), 14–29.
Kitchin, R., & Lauriault, T. P. (2015). Small data in the era of big data. In GeoJournal, 80(4), 463–475.
Kleinberg, Jon M.; Ludwig, Jens; Mullainathan, Sendhill (2016): A Guide to Solving Social Problems
with Machine Learning (Harvard Business Review). Available online at https://​hbr.​org/​2016/​12/a-​
guide-​to-​solvi​ng-​social-​probl​ems-​with-​machi​ne-​learn​ing, checked on 12/1/2017.
Köbis, Nils; Bonnefon, Jean-François; Rahwan, Iyad (2021): Bad machines corrupt good morals. In Nat
Hum Behav.
Kohlberg, Lawrence; Levine, Charles; Hewer, Alexandra (1983): Moral stages. A current formulation and
a response to critics. Basel: Karger.
Koopman, P., & Wagner, M. (2017). Autonomous vehicle safety: an interdisciplinary challenge. In IEEE
Intell. Transport. Syst. Mag., 9(1), 90–96.
Kosinski, M., Bachrach, Y., Kohli, P., Stillwell, D., & Graepel, T. (2014). Manifestations of user per-
sonality in website choice and behaviour on online social networks. In Machine Learning, 95(3),
357–380.
Kosinski, M., Stillwell, D., & Graepel, T. (2013). Private traits and attributes are predictable from digital
records of human behavior. In Proceedings of the National Academy of Sciences of the United
States of America, 110(15), 5802–5805.
Kuss, Daria J.; Griffiths, Mark D. (2017): Social Networking Sites and Addiction. Ten Lessons Learned.
In International journal of environmental research and public health 14 (3).
Kwapisz, J. R., Weiss, G. M., & Moore, S. A. (2011). Activity recognition using cell phone accelerom-
eters. In SIGKDD Explor. Newsl., 12(2), 74–82.

13
590
T. Hagendorff

Lerner, R. M. (Ed.). (2015). Handbook of child psychology and developmental science. Hoboken, New
Jersey: Wiley.
Lessig, Lawrence (2006): Code. Version 2.0. New York: Basic Books.
Lewis, K. (2015). Three fallacies of digital footprints. In Big Data & Society, 2(2), 1–4.
Lischka, Konrad; Stöcker, Christian (2017): Digitale Öffentlichkeit. Wie algorithmische Prozesse den
gesellschaftlichen Diskurs beeinflussen. Arbeitspapier. Gütersloh: Bertelsmann Stiftung, pp. 1–88.
Loevinger, J. (1997). Stages of Personality Development. In R. Hogan, J. Johnson, & S. Briggs (Eds.),
Handbook of Personality Psychology (pp. 199–208). Elsevier.
Los, Maria (2006): Looking into the future: surveillance, globalization and the totalitarian potential. In
David Lyon (Ed.): Theorizing Surveillance. The panopticon and beyond. Cullompton: Willian Pub-
lishing, pp. 69–94.
Luhmann, N. (1995). Social Systems. Stanford University Press.
Lyon, David (2003): Surveillance as social sorting. Computer codes and mobile bodies. In David Lyon
(Ed.): Surveillance as Social Sorting. Privacy, risk, and digital discrimination. London: Routledge,
pp. 13–30.
Matz, S. C., & Netzer, O. (2017). Using Big Data as a window into consumers’ psychology. In Current
Opinion in Behavioral Sciences, 18, 7–12.
Matzner, Tobias (2016): Beyond data as representation. The performativity of Big Data in surveillance.
In Surveillance & Society 14 (2), pp. 197–210.
Mayer, N., Ilg, E., Fischer, P., Hazirbas, C., Cremers, D., Dosovitskiy, A., & Brox, T. (2018). What makes
good synthetic training data for learning disparity and optical flow estimation? In Int J Comput vis,
126(9), 942–960.
Mayer-Schönberger, Viktor; Cukier, Kenneth (2013): Big Data. A Revolution That Will Transform How
We Live, Work, and Think. New York: Eamon Dolan.
McCloud, R. F., Okechukwu, C. A., Sorensen, G., & Viswanath, K. (2016). Entertainment or health?
exploring the internet usage patterns of the urban poor: a secondary analysis of a randomized con-
trolled trial. In Journal of Medical Internet Research, 18(3), 1–12.
McCrae, R. R., & John, O. P. (1992). An introduction to the five-factor model and its applications. In
Journal of Personality, 60(2), 175–215.
Mittelstadt, Brent; Russell, Chris; Wachter, Sandra (2019): Explaining Explanations in AI. In Proceed-
ings of the Conference on Fairness, Accountability, and Transparency - FAT* ’19, pp. 1–10.
Mjolsness, Eric; DeCoste, Dennis (2001): Machine learning for science. State of the art and future pros-
pects. In Science 293 (5537), pp. 2051–2055.
Mossberger, Karen; Tolbert, Caroline J.; Stansbury, Mary (2003): Virtual inequality. Beyond the digital
divide. Washington, D.C.: Georgetown University Press.
Mühlhoff, Rainer (2019): Human-aided artificial intelligence: Or, how to run large computations in
human brains? Toward a media sociology of machine learning. In New Media & Society, 1–17.
Mühlhoff, Rainer (2021): Predictive Privacy: Towards an Applied Ethics of Data Analytics. In SSRN
Journal, pp. 1–24.
Myers, Isabel Briggs; Myers, Peter B. (1995): Gifts Differing. Understanding Personality Type. Palo
Alto: Davies-Black.
Ng, Andrew (2020): Facebook Likes Extreme Content (The Batch). Available online at https://​blog.​deepl​
earni​ng.​ai/​blog/​the-​batch-​faceb​ooks-​unruly-​algor ​ithm-​ai-​that-​does-​the-​dishes-​new-​life-​for-​old-​
data-​models-​that-​take-​short​cuts-​yolo-​retur​ns, checked on 6/19/2020.
Olteanu, A., Castillo, C., Diaz, F., & Kıcıman, E. (2019). Social data: biases, methodological pitfalls, and
ethical boundaries. In Front. Big Data, 2, 1–33.
O’Neil, Cathy (2016): Weapons of Math Destruction. How Big Data Increases Inequality and Threatens
Democracy. New York: Crown Publishers.
Pang, Bo., & Lee, L. (2008). Opinion mining and sentiment analysis. In FNT in Information Retrieval,
2(1–2), 1–135.
Patel, T. (2006). Comparing the usefulness of conventional and recent personality assessment tools. In
Global Business Review, 7(2), 195–218.
Paul, B., & Baltes; David L. Featherman; Richard M. Lerner; Orville Gilbert Brim; Marion Perlmutter, .
(1978). Life Span Development and Behavior. Academic Press Inc.
Perrons, R. K., & McAuley, D. (2015). The case for “n«all”: why the big data revolution will probably
happen differently in the mining sector. In Resources Policy, 46, 234–238.
Picard, R. W. (1997). Affective computing. MIT Press.

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 591

Rahwan, I., Cebrian, M., Obradovich, N., Bongard, J., Bonnefon, J.-F., Breazeal, C., et  al. (2019).
Machine behaviour. In Nature, 568(7753), 477–486.
Raji, Inioluwa Deborah; Smart, Andrew; White, Rebecca N.; Mitchell, Margaret; Gebru, Timnit; Hutch-
inson, Ben et al. (2020): Closing the AI Accountability Gap: Defining an End-to-End Framework
for Internal Algorithmic Auditing. In arXiv, pp. 1–12.
Rawls, J. (1999). A Theory of Justice. Oxford University Press.
Roßnagel, A. (2007). Datenschutz in einem informatisierten Alltag. Gutachten im Auftrag der
Friedrich-Ebert-Stiftung.
Samarati, Pierangela; Sweeney, Latanya (1998): Protecting Privacy when Disclosing Information: k-Ano-
nymity and Its Enforcement through Generalization and Suppression. In Technical Report SRI,
pp. 1–19.
Samitsch, Christoph (2015): Data Quality and its Impacts on Decision-Making. How Managers can ben-
efit from Good Data. Wiesbaden: Springer.
Sartorius, R. (Ed.). (1983). Paternalism. University of Minnesota Press.
Schneier, Bruce (2015): Data and Goliath. The Hidden Battles to Collect Your Data and Control Your
World. New York: W. W. Norton & Company.
Schulz, Eric; Dayan, Peter (2020): Computational Psychiatry for Computers. In iScience 23 (12),
p. 101772.
Schulze, Gerhard (1996): Die Erlebnis-Gesellschaft. Kultursoziologie der Gegenwart. Frankfurt am
Main: Campus Verlag.
Schwartz, H., Andrew; Eichstaedt, Johannes C., Kern, Margaret L., Dziurzynski, Lukasz; Ramones,
Stephanie M., Agrawal, Megha, , et  al. (2013). Personality, gender, and age in the language of
social media: the open-vocabulary approach. In PloS One, 8(9), 1–16.
Segalin, Cristina; Celli, Fabio; Polonio, Luca; Kosinski, Michal; Stillwell, David; Sebe, Nicu et  al.
(2017): What your Facebook Profile Picture Reveals about your Personality. In Qiong Liu, Rainer
Lienhart, Haohong Wang, Sheng-Wei "Kuan-Ta" Chen, Susanne Boll, Phoebe Chen et al. (Eds.):
Proceedings of the 2017 ACM on Multimedia Conference - MM ’17. New York: ACM Press,
pp. 460–468.
Sheng, Emily; Chang, Kai-Wei; Natarajan, Premkumar; Peng, Nanyun (2019): The Woman Worked as
a Babysitter: On Biases in Language Generation. In Kentaro Inui, Jing Jiang, Vincent Ng, Xiao-
jun Wan (Eds.): Proceedings of the 2019 Conference on Empirical Methods in Natural Language
Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-
IJCNLP). Stroudsburg, PA: Association for Computational Linguistics, pp. 3405–3410.
Shorten, C., & Khoshgoftaar, T. M. (2019). A survey on image data augmentation for deep learning. In J
Big Data, 6(1), 1–48.
Silvervarg, A., Raukola, K., Haake, M., & Gulz, A., et al. (2012). The Effect of Visual Gender on Abuse
in Conversation with ECAs. In D. Hutchison, T. Kanade, J. Kittler, J. M. Kleinberg, F. Mattern, &
J. C. Mitchell (Eds.), Intelligent Virtual Agents (pp. 153–160). Springer.
Sloterdijk, Peter (2009): Du mußt dein Leben ändern. Über Anthropotechnik. Frankfurt a.M: Suhrkamp.
Solaiman, Irene; Clark, Jack; Brundage, Miles (2019): GPT-2: 1.5B Release. OpenAI. Available online at
https://​openai.​com/​blog/​gpt-2-​1-​5b-​relea​se/, checked on 11/13/2019.
Spera, C. (2005). A review of the relationship among parenting practices, parenting styles, and adolescent
school achievement. Educational Psychology Review, 17(2), 125–146.
Spindler, M., Booz, S., Gieseler, H., Runschke, S., Wydra, S., & Zinsmaier, J. (2020). How to achieve
integration? In B. Gransche & A. Manzeschke (Eds.), Das geteilte Ganze (pp. 213–239). Springer
Fachmedien Wiesbaden.
Stieglitz, Stefan; Dang-Xuan, Linh (2012): Impact And Diffusion Of Sentiment In Public Communication
On Facebook. In ECIS 2012 Proceedings 98, pp. 1–12.
Stieglitz, Stefan; Dang-Xuan, Linh (2013): Emotions and Information Diffusion in Social Media. Senti-
ment of Microblogs and Sharing Behavior. In Journal of Management Information Systems 29 (4),
pp. 217–248.
Sullivan, P. M., & Knutson, J. F. (2000). Maltreatment and disabilities: A population-based epidemiologi-
cal study. Child Abuse & Neglect, 24(10), 1257–1273.
Suresh, Harini; Guttag, John V. (2020): A Framework for Understanding Unintended Consequences of
Machine Learning. In arXiv, pp. 1–10.
Tellegen, Auke; Waller, Niels G. (2008): Exploring Personality Through Test Construction: Development
of the Multidimensional Personality Questionnaire. In Gregory J. Boyle, Gerald Matthews, Don

13
592
T. Hagendorff

Saklofske (Eds.): The SAGE Handbook of Personality Theory and Assessment: Volume 2. Person-
ality Measurement and Testing. London: SAGE Publications Ltd, pp. 261–292.
Brundage, Miles; Avin, Shahar; Wang, Jasmine; Belfield, Haydn; Krueger, Gretchen; Hadfield, Gillian
et al. (2020): Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims.
In arXiv, pp. 1–80.
Tsutsui, Satoshi; Kerola, Tommi; Saito, Shunta; Crandall, David J. (2018): Minimizing Supervision for
Free-space Segmentation. In arXiv, pp. 1–10.
Turow, Joseph (2012): The Daily You. How the New Advertising Industry Is Defining Your Identity and
Your Worth. New Haven: Yale University Press.
Underwood, Mimi (2015): Updating Our Search Quality Rating Guidelines (Google Blog). Available
online at https://​webma​sters.​googl​eblog.​com/​2015/​11/​updat​ing-​our-​search-​quali​ty-​rating.​html,
checked on 5/29/2020.
Brundage, Miles; Avin, Shahar; Clark, Jack; Toner, Helen; Eckersley, Peter; Garfinkel, Ben et al. (2018):
The Malicious Use of Artificial Intelligence. Forecasting, Prevention, and Mitigation. In arXiv,
pp. 1–101.
van Dijck, J. (2014). Datafication, dataism and dataveillance: big data between scientific paradigm and
ideology. In Surveillance & Society, 12(2), 197–208.
van Ly, Minh; Martin, Sujitha; Trivedi, Mohan M. (2013): Driver classification and driving style recogni-
tion using inertial sensors. In : 2013 IEEE Intelligent Vehicles Symposium (IV). Piscataway: IEEE,
pp. 1040–1045
Vaughn, M. G., Define, R. S., Delisi, M., Perron, B. E., Beaver, K. M., Fu, Q., & Howard, M. O. (2011).
Sociodemographic, behavioral, and substance use correlates of reckless driving in the United
States: findings from a national Sample. In Journal of Psychiatric Research, 45(3), 347–353.
Veale, Michael; Binns, Reuben (2017): Fairer machine learning in the real world. Mitigating discrimina-
tion without collecting sensitive data. In Big Data & Society 4 (2), 1–17.
Vester, Michael (2001): Soziale Milieus im gesellschaftlichen Strukturwandel. Zwischen Integration und
Ausgrenzung. Frankfurt: Suhrkamp.
Wallach, Wendell; Allen, Colin (2009): Moral Machines. Teaching Robots Right from Wrong. New York:
Oxford University Press.
Wang, Richard Y.; Strong, Diane M. (1996): Beyond Accuracy. What Data Quality Means to Data Con-
sumers. In Journal of Management Information Systems 12 (4), pp. 5–33.
Wang, X., Huang, K., & Yang, Li. (2019). Effects of socio-demographic, personality and mental health
factors on traffic violations in Chinese bus drivers. In Psychology, Health & Medicine, 24(7),
890–900.
Wayne, N. L., & Miller, G. A. (2018). Impact of gender, organized athletics, and video gaming on driving
skills in novice drivers. In PloS One, 13(1), 1–12.
Wedel, M., & Kannan, P. K. (2016). Marketing analytics for data-rich environments. In Journal of Mar-
keting, 80(6), 97–121.
Wendehorst, Christiane; Woopen, Christiane; Haberer, Johanna; Heckmann, Dirk; Hansen, Marit; Kelber,
Ulrich et al. (2019): Opinion of the Date Ethics Commission. Berlin: Data Ethics Commission of
the Federal Government, pp. 1–240. Available online at https://​www.​bmjv.​de/​Share​dDocs/​Downl​
oads/​DE/​Themen/​Fokus​themen/​Gutac​hten_​DEK_​EN.​pdf?__​blob=​publi​catio​nFile​&v=2, checked
on 7/4/2021.
West, Mark; Kraut, Rebecca; Chew, Han Ei (2019): I’d blush if I could: closing gender divides in digital
skills through education, pp. 1–146.
West, R., & Hall, J. (1997). The role of personality and attitudes in traffic accident risk. In Applied Psy-
chology, 46(3), 253–264.
Westin, A. F. (1967). Privacy and Freedom. Atheneum.
Willke, Helmut (2005): Symbolische Systeme. Grundriss einer soziologischen Theorie. Weilerswist:
Velbrück.
Wolf, M. J., Miller, K., & Grodzinsky, F. S. (2017). Why we should have seen that coming. In SIGCAS
Comput. Soc., 47(3), 54–64.
Woodall, P., Oberhofer, M., & Borek, A. (2014). A classification of data quality assessment and improve-
ment methods. In IJIQ, 3(4), 298–321.
Xu, C., Ding, Z., Wang, C., & Li, Z. (2019). Statistical analysis of the patterns and characteristics of con-
nected and autonomous vehicle involved crashes. In Journal of Safety Research, 71, 41–47.
Young, M. D. (1994). The Rise of the Meritocracy. Transaction.

13
Linking Human And Machine Behavior: A New Approach to Evaluate… 593

Yu, Pu., & Grossklags, J. (2016). Towards a model on the factors influencing social app users’ valuation
of interdependent privacy. In Proceedings on Privacy Enhancing Technologies, 2, 61–81.
Zeng, Zhihong; Pantic, Maja; Roisman, Glenn I.; Huang, Thomas S. (2009): A survey of affect recog-
nition methods: audio, visual, and spontaneous expressions. In IEEE Trans. Pattern Anal. Mach.
Intell. 31 (1), pp. 39–58.
Zhuang, Fuzhen; Qi, Zhiyuan; Duan, Keyu; Xi, Dongbo; Zhu, Yongchun; Zhu, Hengshu et al. (2020): A
Comprehensive Survey on Transfer Learning. In arXiv, pp. 1–31.
Zillien, N., & Hargittai, E. (2009). Digital distinction: status-specific types of internet usage. In Social
Science Quarterly, 90(2), 274–291.
Zuboff, S. (2015). Big other: Surveillance capitalism and the prospects of an information civilization. In
Journal of Information Technology, 30, 75–89.

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.

13