Relative Strengths of Teachers and Smart Machines: Towards an

Augmented Task Sharing

Michael Burkhard, Sabine Seufert and Josef Guggemos

Institute for Educational Management and Technologies, University of St. Gallen,

St. Jakob-Strasse 21, 9000 St. Gallen, Switzerland

Keywords: Education, Human Augmentation, Comparative Advantage, Role of the Teacher, Smart Machine,

Social Robot, Chatbot.

Abstract: In education, smart machines (e.g., chatbots or social robots) have the potential to support teachers in the

classroom in order to improve the quality of teaching. From a teacher's point of view, smart machines also

pose a challenge because the presence of smart machines in the classroom questions traditional teacher and

student roles. This paper presents a theoretical basis for the use of smart machines in education. It describes

the relative strengths of teachers and smart machines and presents them in a framework, which makes a pro-

posal for an augmented task sharing. In light of human augmentation, the framework proposes ways in which

teachers can position themselves with regard to smart machines in a complementary and mutually reinforcing

way. It also has implications for knowledge that is necessary for teachers to play an active role in the digital

transformation.

1 INTRODUCTION

The society, economy, and the labor market are on the

threshold of a major transition phase. Widely used la-

bels for this phase are: The fourth industrial revolu-

tion (Braga et al., 2019), the second machine age

(Brynjolfsson & McAfee, 2014), the second wave of

digitalization (Wahlster, 2017), artificial intelligence

(AI) revolution (Makridakis, 2017), and globotics

(globalization and robotics) (Baldwin, 2020). Tech-

nological developments in robotics combined with

machine learning and AI underscore the importance

of a better understanding of the human-machine rela-

tionship, as humans and machines may become part-

ners in learning and problem solving (Brynjolfsson &

McAfee, 2014; Jarrahi, 2018). Humans and smart ma-

chines engage in task sharing and combine their indi-

vidual strengths.

These technological developments also have an

influence on classrooms. Teachers become

increasingly part of a digital classroom ecosystem.

Such smart classrooms are equipped with tools that

facilitate the transfer of knowledge (e.g., by means of

more efficient communication or automated

assessment/feedback), with the goal to enhance the

teaching and learning experience (Saini & Goel,

2019, pp. 1-2).

As part of a digital classroom ecosystem, a smart

machine can be defined as a cognitive computer

system that can, to a certain extent, make decisions

and solve problems without the help of a human being

(Pereira, 2019). This is achieved by advanced

technology (e.g., AI, machine learning), which

enables the machine to process a large amount of data

and make decisions based on these data.

Chatbots (e.g., Apple´s Siri) or social robots can

be regarded as important manifestation of smart

machines, provided that these smart machines are

capable of learning from the environment and build

on capabilities based on that knowledge (Pereira,

2019).

Smart machines are increasingly used in everyday

life due to advances in sensor and actuator

technology. During the last ten years, the use of smart

machines has been increasingly extended to the field

of education, starting with the use as an aid in STEM

education (Belpaeme et al., 2018). “Socially

conscious” robots interact for example with children

in language learning classes (Van den Berghe,

Verhagen, Oudgenoeg-Paz, van der Ven & Leseman,

2019). According to Reich-Stiebert, Eyssel and

Hohnemann (2019, p. 5) such robots can be used as

assistants to teachers or personal tutors for students:

“provide information on specific topics, query

Burkhard, M., Seufert, S. and Guggemos, J.

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing.

DOI: 10.5220/0010370300730083

In Proceedings of the 13th International Conference on Computer Supported Education (CSEDU 2021) - Volume 1, pages 73-83

ISBN: 978-989-758-502-9

learned lessons, give advice to the learning process,

correct errors, or provide feedback on students’

progress” (Reich-Stiebert et al., 2019, p.5).

Unlike the digital classroom ecosystem (e.g.,

projectors, cameras, interactive white boards), smart

machines are perceived as more than just a tool. Due

to their nature, they act as someone (personality) and

not as something (tool).

A teacher has many different tasks to perform.

These are for example to plan lessons, to teach, to

coach, to create assignment and homeworks, to

conduct and correct exams, to manage the classroom,

and to activate students.

From a teacher's point of view, the additional

presence of a smart machine could be beneficial as

the smart machine can engage in task sharing and

take over selected duties of the teacher. However,

smart machines (respectively the inherent technology

of AI) have also the potential to replace white-collar

jobs (see e.g, Baldwin 2020, p. 9). Hence, the smart

machine could also be perceived as a threat, because

the presence of the smart machine in the classroom

challenges the traditional role of teachers and

students.

At the moment, there is a gap between the

available technological capabilities and their

utilization for educational purposes (Luan et. al.,

2020, p. 3). Even though the education industry has

developed various AI applications, they may not be

guided by theoretical frameworks and research

findings from psychology of learning and teaching

(Luan et. al., 2020, p. 3). There seems to be a disparity

between the technology readiness and its application

in education (Macfadyen, 2017).

To tackle this issue, it might be important to gain

a better understanding of the relative strengths of

teachers and smart machines. Afterwards, based on

the theory of comparative advantages (Ricardo, 1891;

Ruffin, 2002; Landsburg, n.d.), ways could be

pointed out in which teachers can position themselves

in relation to smart machines in a complementary and

mutually reinforcing way.

In light of the identified research desideratum, the

following research questions should be addressed:

 What are the relative strengths of teachers and

smart machines within the classroom?

 How can both parties engage in an augmented,

mutually reinforcing way of task sharing?

The objectives of the paper at hand are therefore

twofold:

 Elaboration of the theoretical foundations for

the use of smart machines in education, in order

to investigate underlying assumptions, goals,

methods, and empirical results for the design

and evaluation of teaching;

 Development of a conceptual framework from

the teacher´s perspective on augmentation

strategies of teachers in relation to smart ma-

chines.

From a theoretical point of view, our conceptual

framework can serve as starting point for future

empirical research, as it highlights important concepts

and variables related to the relative strengths of

teachers and smart machines.

From a practical standpoint, our conceptual

framework might be useful for designing use cases; it

could serve as a guideline in the implementation

process of new technologies. Overall, the conceptual

framework at hand might act as a stepping stone for

coming researchers who might uncover further

potential of the technology in more detail, e.g., how

to ensure a smooth adoption of social robots, as a

concrete manifestation of smart machines, in

education.

To this end, we lay the foundation for our

framework in section 2 and 3. Since comparative

advantages of smart machines may depend on the

environment and context of use, we will point out the

relationship of smart machine and the digital classroom

ecosystem in section 2. Section 3 discusses relative-

strength profiles of teachers and smart machines based

on the theory of comparative advantages and evaluates

them with regard to specific teaching tasks. Section 4

lays out our own extended framework, and section 5

concludes with some final remarks.

2 SMART MACHINES AS PART

OF A CLASSROOM

ECOSYSTEM

According to Floridi (2016), we are in transition to a

new era in which we will become increasingly de-

pendent from our own technical achievements. ICT is

not only used to record and transmit data, but also to

process it more and more autonomously.

Floridi (2013, pp. 6-7) coined the term “in-

fosphere”, i.e., an information environment compara-

ble to, but different from, cyber space, which is be-

coming increasingly blurred with our everyday life.

The infosphere is constituted by all informational en-

tities (biological as well as digital agents/smart arte-

facts). A digital classroom ecosystem can also be seen

as such a form of an infosphere and is often referred

to as smart classroom (see e.g., Saini & Goel, 2019).

CSEDU 2021 - 13th International Conference on Computer Supported Education

Source: Based on Lehmann & Rossi (2019, p. 36) and own contributions.

Figure 1: Changes in interaction due to smart machines.

Biological agents (teachers and students) as well as

digital artefacts (e.g., tools such as interactive white-

boards, laptops, smartphones) interact in an ecosys-

tem according to a syllabus.

Digital classroom ecosystems have the potential to

facilitate the transfer of knowledge from teacher to stu-

dents in various ways (Saini & Goel, 2019). It can sup-

port the teacher in content creation, content presenta-

tion, and content distribution (Saini & Goel, 2019, pp.

6-12) promote interaction between different biological

agents (Saini & Goel, 2019, pp. 12-14) and provide au-

tomated assessment and feedback as well as some

background functions (e.g., temperature control inside

the classroom) (Saini & Goel, 2019, pp. 15-20). Due to

the nature of an infosphere, the digital classroom eco-

system can be seen as an advanced tool (like a car) that

helps the teachers to better achieve their goals. It sup-

ports teachers to get from A to B more quickly, but

teachers still have to steer and to drive.

In contrast, smart machines rather play a collabo-

rative role because they are perceived as a form of

digital personality and to a certain extent can make

decisions and solve problems without the help of the

teacher. Smart machines are not just a tool (some-

thing), but someone, who in certain cases could also

be sitting in the driver’s seat. This leads to changes in

the classroom interaction as Figure 1 illustrates.

In the context of smart machines, Lehmann and

Rossi (2019) propose an enactive robot assisted di-

dactics (ERAD) approach, where robots act as inter-

mediaries and catalysts between teacher, students,

and context (see Figure 1). Smart machines can per-

form such a role because they generate attention and

expectations in both teachers and students, which en-

ables them to influence and adapt the behaviour of

their human counterpart.

The presence of the smart machine in the class-

room changes the situation in teaching. From the

teacher's point of view, new questions arise. Some of

these questions could be: What role does the smart

machine play in relation to the learners? For which

parts of the curriculum is the smart machine suitable

to provide support? What role do I play as a teacher

when the smart machine is suddenly present? These

questions can cause stress or even lead to anxiety

about being replaced by the robot.

Since the smart machine is perceived as a form of

digital personality, possible roles that the smart ma-

chine can play are important. In education, according

to Sharkey (2016) four main roles exist:

1. Smart machines as teachers (e.g., to take over

selected teacher duties in the classroom);

2. Smart machines as companions and peers

(e.g., to work collaboratively with students);

3. Smart machines as care-eliciting companions

(e.g., supporting students with disabilities);

and

4. Smart machines as telepresence teachers

(e.g., online teaching through digital technol-

ogies along the lines of teachers in distance

education).

On the one hand, these different role models show

that smart machines (respectively the inherent tech-

nology of AI) tend to contain a disruptive potential,

because the machine is perceived on a par with the

teacher. Unlike digital classroom ecosystems in gen-

eral, the inherent role of the smart machine confers a

certain authority that could challenge the authority

and competence of the human teacher. Table 1 com-

pares smart machines and the digital classroom eco-

system to clearly point out the differences.

Smart machines could offer a learning experience

tailored to the learner, support and challenge students,

and free up precious time for human teachers through

ways currently unavailable in our educational environ-

ments (Belpaeme et. al, 2018, p. 7). In addition, as an

adapter between the digital and analogue world, smart

machines would be ideally suited to manage the digital

classroom ecosystem according to teachers’ needs.

Syllabus

Tea che r

Student

Tools

ClassroomEcosystemwithoutaSmartMachine

Smartmachine

Tea che r

Student

Syllabus

Tools

ClassroomEcosystemwithaSmartMachine

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing

Table 1: Comparison between a smart machine and a digital classroom ecosystem.

Factors Characteristics of a smart machine Characteristics of a di

ital classroom ecos

stem

Perception Digital personality,

igital agent Tool, digital environment, artefact

ole Someone

(

can also be in co-role or lead

)

Somethin

(

an advanced tool

)

Representation Generic chatbots, social robots

The connected eco-system of technology inside the

classroom (e.g. interactive whiteboards, projectors,

cameras,

rinters, smart-

hones

)

Underlying

technology

AI, machine learning, ICT Information and communication technology (ICT)

Nature of work

Make decisions and solve problems without

the help of a human being

Data collection- and decision-support-system

Disruptive

otential

Disruptive (potential to substitute the

teacher

)

Incremental (human teacher required)

Both chatbots and social robots are manifestations

of smart machines. In addition to chatbots, social ro-

bots also have a physical presence. A new field of re-

search is currently emerging: Human-Robot-Interac-

tions or social robots in education (Belpaeme et al.,

2018; Byrne, Rossi & Doolan, 2017; Chua & Chew,

2015; Flynn, 2017). The emerging use of robots is

changing human augmentation, as these smart ma-

chines have a physical presence. In the field of hu-

man-computer interactions, a robot is not only a com-

puter-based machine, but also a physical and autono-

mous agent, whose physical form and degree of au-

tonomy influences the relationship to humans

(Thimm et al., 2019).

In the field of education, there are several studies

where robots have been used to teach groups of sub-

stantial size (e.g., Abildgaard & Scharfe, 2012;

Cooney & Leister, 2019; Guggemos, Seufert &

Sonderegger, 2020; Masuta et al., 2018) but also to

teach smaller workshop-like (e.g., Bolea, Grau, &

Sanfeliu, 2016) or even one-on-one interactions (e.g.,

Gao, Barendregt, Obaid, & Castellano, 2018).

According to Belpaeme et al. (2018, p. 7), smart

machines such as social robots have the potential to be-

come part of the educational infrastructure, just as pa-

per, white boards or computer tablets. In their meta-

analysis they gathered results from a wide range of

countries and took different robot types and ap-

proaches into account. They conclude that robots show

great promise when teaching restricted topics, with ef-

fect sizes on cognitive outcomes almost matching

those of human tutoring (Belpaeme et al., 2018, p. 7).

If smart machines may become part of the educa-

tional infrastructure in the future, ways must be found

to enable smart machines and teachers to collaborate in

the classroom for mutual benefit. With the goal of cre-

ating a meaningful task sharing in the classroom, we

will therefore take a closer look at the relative strengths

of teacher and smart machine in the next section.

3 COMPARATIVE STRENGTHS

OF TEACHERS AND SMART

MACHINES

Among the many effects, digitalization will have on

our way of working and living, the augmentation of

human skills is the most central (Davenport & Kirby,

2016). Davenport and Kirby (2016) draw attention to

the mutual complementation and task sharing that

they call “augmentation: People and computers sup-

ported each other in the fulfilment of tasks” (p. 2).

According to Jarrahi (2018) augmentation can be un-

derstood as a “Human-AI symbiosis” meaning that

interactions between humans and AI can make both

parties smarter over time (p. 583).

Figure 2 illustrates this relationship. On the one

hand, people have to train machines to perform cer-

tain tasks. They have to explain the results of those

tasks to other stakeholders and ensure the responsible

use of machines. On the other hand, smart machines

help people by enhancing their cognitive strengths,

relieving them from repetitive tasks, and expanding

their physical abilities.

To investigate the relative strengths of the teacher

and the smart machine, we use as a foundation the

theory of comparative advantage, which originates

from the field of economics (Ruffin, 2002; Lands-

burg, n.d.). Ricardo (1891) was first able to show with

his theory, why two countries A and B engage in

trade, even if one country is in absolute terms supe-

rior to the other regarding the production of all goods

in the economy. He was able to explain, why coun-

tries specialize on the production of certain goods and

trade them. He showed that not the absolute ad-

vantage matters (being better at producing all goods),

but the relative advantage instead (having lower op-

portunity costs).

CSEDU 2021 - 13th International Conference on Computer Supported Education

Source: Own representation based on Wilson & Daugherty (2018).

Figure 2: Human Augmentation.

When we apply the concept of comparative ad-

vantages to the classroom, teacher and smart machine

can be seen as countries A and B. In the classroom dif-

ferent tasks have to be carried out (production of

goods). For example, two of these tasks could be to

provide feedback to homework or to individually sup-

port students. To illustrate the comparative advantage,

we assume the task times of teacher and smart machine

depicted in Table 2. Differences in task quality are im-

plicitly reflected in longer task times.

Table 2: Comparison of the task time (absolute).

Task-time needed Teache

Smart machine

Individual Coachin

10 min 15 min

Provide Feedbac

5 min 15 min

As it can be seen in Table 2, in absolute terms, the

teacher is better in both tasks coaching and feedback

(lower task times). The question is: Can it be benefi-

cial for the teacher to shift tasks to the smart machine?

According to the theory of comparative ad-

vantages it can, because not the absolute but the rela-

tive advantages matter. Table 3 shows the opportunity

costs for our scenario.

Table 3: Comparison of the opportunity costs (relative).

ortunit

costs Teache

Smart machine

Individual Coaching

10/5 = 2

Feedbac

15/15 = 1

Feedbac

Provide Feedback

5/10 = 0.5

Coachin

15/15 = 1

Coachin

As the smart machine is equally fast in both tasks,

for every unit of coaching, the smart machine cannot

produce a unit of feedback. Hence their opportunity

costs are 1 for both coaching and feedback.

However, as the teacher is much faster providing

feedback than coaching, the opportunity costs for

coaching are very high, as for each coaching he or she

cannot give two units of feedback.

Each party (teacher and smart machine) should do

the tasks where they have lower opportunity costs com-

parted to their counterpart. In the example at hand, the

teacher will specialize on providing feedback (0.5 < 1)

and the smart machine will do coaching (1<2).

Our example shows, that smart machines can be

useful even if they are inferior to humans in absolute

terms. On a more general level, smart machines have

comparative advantages over the teacher in certain

fields. Hence, it is beneficial that they take over spe-

cific tasks for the teacher. This means that a given set

of tasks can be carried out in less time (costs) or

within a given time (costs), the number of carried

tasks (quality) can be increased.

As it has been shown, the crucial point is the rela-

tive strengths of teachers and smart machines. Jarrahi

(2018) created relative strength-profiles of humans

and AI regarding their core skill set along three di-

mensions: uncertainty, complexity and equivocality

(Jarrahi, 2018, p. 583), see Figure 3.

When assessing the threat posed by technology to

a particular profession, Latham and Humberd (2018,

p. 12) point out that it is important to look at the core

skill set, but also at how the value of the core skill set

is delivered (value form). Latham and Humbert

(2018, p. 13) grounded the value form in consumer

preferences, task diversity and wage differences, on

the basis of which we created the three dimensions

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing

preferences, variety and attractivity. Figure 3 summa-

rizes the created relative strength-profile of teachers

and smart machines along the two categories core

skill set and value form as well as the six dimensions

uncertainty, complexity, equivocality, preferences,

variety and attractivity.

Jarrahi (2018, pp. 580-581) characterizes uncer-

tainty as a lack of information about all alternatives

or their consequences, which makes interpreting a sit-

uation and making a decision more difﬁcult. He ar-

gues that for situations, which there is no precedent,

an intuitive style of decision making may be more

helpful. According to Jarrahi (2018, pp. 580-581) in

the dimension of uncertainty, humans have a relative

advantage over AI due to their ability of intuitive de-

cision making (e.g., Harteis & Billett, 2013). Smart

machines can still help to reduce uncertainty by

providing access to real time information, but as ma-

chines are mostly incapable of capturing the inner

logic and subconscious patterns of human intuition,

humans tend to keep their comparative advantage in

situations that require holistic and visionary thinking

(Jarrahi, 2018, p. 581).

In the classroom, uncertainty may occur through

different channels. On the one hand, students may ask

surprising questions or give inputs, that require some

forms of intuitive thinking or creativity in order to an-

swer the question. On the other hand, classroom dy-

namics itself are to a certain degree unpredictable and

uncertain as students are individuals with their own

needs. Students do not behave the same way every

day, and sometimes they may not even show up. How

to react to these situations requires intuition and can-

not be solved by a fixed rule alone.

Complexity is characterized by an abundance of

elements or variables, that demand the processing of

masses of information. AI has a comparative ad-

vantage in handling complexity due to their ability of

collecting, curating, processing, and analyzing large

amounts of data (Jarrahi, 2018, p. 581).

In the classroom, complexity increases with the

number of students as the same assignments, exer-

cises and exams are conducted for more people. With

more students, it gets more difficult to keep an over-

view over the learning success of each student. Espe-

cially in large classes, smart machines can be a valu-

able research if they support the teacher in providing

automated feedback for homework and exams.

Equivocality is characterized by the presence of

several simultaneous but divergent interpretations of a

decision domain and often occurs due to the conﬂicting

interests of stakeholders, customers, and policy makers

(Jarrahi, 2018, p. 581). It means, that there is not al-

ways one objective solution to a problem, but multiple

different and subjective views about an issue. Even

though smart machines may be able to analyse senti-

ments and represent diverse interpretations, humans

have a comparative advantage, when it comes to han-

dling equivocality as they are better in negotiating and

coalition building (Jarrahi, 2018, p. 582).

In the classroom, equivocality may occur due to

different circumstances. One the one hand, the sylla-

buses of certain school subjects may be more subjec-

tive than others. While subjects like accounting or

mathematics provide clear guidance on “true” and

“false”, this line is more difficult to draw in subjects

such as history or literature. On the other hand, stu-

dents often also have different opinions and there is a

need for a teacher who can work out a common con-

sent during discussions.

A smart machine and a human teacher are very

different by nature. Hence, for certain tasks it will de-

pend simply on the preferences

of the students, who

they address with their problems.

In the classroom, preferences depend primarily on

social norms and informal social rules between hu-

mans. In human conversation, there are informal rules

that have to be followed (e.g., be friendly), which are

time consuming and can make communication

Source: (1) Latham & Humberd (2018), (2) Jarrahi (2018) and own contributions.

Figure 3: Relative strength-profiles of teachers and smart machines.

Uncertainty

(2)

Makeswift,

intuitive

decisionsinthe

faceofthe

unknown.

Provideaccessto

realtime

information(e.g.

Definitions,

statistics).

Complexity

(2)

Keepanoverview,

evaluate

recommendations

ofthesmart

machine,plan

nextsteps.

Collect,curate,

process,and

analyzedata(e.g.

homeworks,

assignments).

Equivocality

(2)

Negotiate,build

consensus,and

rallysupport(e.g.

mediatebetween

students,guide

discussions).

Analyze

sentimentsand

representdiverse

interpretations

(e.g.giveinputs

fordiscussions).

Preferences

Thestudent

preferstoreceive

theservicefroma

human(e.g.

receivepositive

feedback).

Thestudent

preferstoreceive

theservicefroma

smartmachine

(e.g.embarrassing

questions)

Variety

Thetaskvaries

fromstudentto

student,which

makesautomation

difficult(e.g.

individual

coaching).

Thetaskis

repetitivefrom

studenttostudent

andcantherefore

beautomated

(e.g. correction of

assignments).

Attractivity

Relativelylow

teacher'swage

costsandhigh

operating costs of

smart machines

decreaseincentives

toautomatetasks.

Relativelyhigh

teachers'wage

costsandlow

operatingcostsof

smartmachines

increaseincentives

toautomatetasks.

Coreskillset

(1)

Valueform

(1)

CSEDU 2021 - 13th International Conference on Computer Supported Education

Table 4: Augmentation strategies of teachers in relation to smart machines.

Augmentation

strategy

Added value to the smart machine

Relative strength

of the teacher

Example

Step In

To train the smart machine and shift

tasks to it.

Uncertainty,

Equivocality

To automate correction of assign-

ments. To decide on the appropriate

content and su

ervise trainin

Step Up

To manage the classroom and its play-

ers, keep an overview, evaluate, decide

on the ethical use of a smart machine.

Uncertainty,

Complexity,

Equivocality

To decide on how to proceed if

homework has not been done. To

decide on appropriate tasks for the

smart machine.

Step Forward

To participate in the content develop-

ment and data analysis of the smart

machine.

Uncertainty,

Equivocality

To develop new teaching content

for a smart machine, to check and

correct for data biases.

Step Aside

To take on tasks that go beyond infor-

mation processing or require tacit

knowledge.

Equivocality,

Preferences

To coach the students, engage with

them in creative problem solving, to

motivate and consult.

Step Narrowly

To perform tasks that cannot be per-

formed well by smart machines (e.g.

non-repetitive tasks).

Variety To maintain the smart machine.

inefficient. Since those rules do not apply to smart ma-

chines, students can ask any question and they do not

have to be afraid of asking a “stupid” question or act-

ing socially inappropriately. Smart machines can also

repeat answers as often as needed (e.g., in language

learning) without getting tired, which makes them a

cooperative learning partner. Teachers may be reluc-

tant to answer the same question several times.

For other tasks, the variety is decisive. If a task

varies from student to student, it will be more difficult

to automate and harder to solve by a smart machine.

However, if a task is repetitive, it can be more easily

automated as the smart machine can be better trained

on it.

In the classroom, variety is task dependent. Espe-

cially the correction of written assignments is repeti-

tive, because the same work steps have to be carried

out for each student. Other tasks like an individual

discussion with a student about his or her research

project differ from student to student and from project

and cannot simply be taken over by a smart machine.

Last but not least, the attractivity to automate

tasks also has an influence if a certain task is shifted

from a human teacher to a smart machine. The attrac-

tivity depends largely on the wage costs of the human

teacher in relation to the operating costs of the smart

machine.

The attractivity depends also on the type of smart

machine. As chatbots have lower operating costs than

social robots, it may be more attractive to offer human

teachers chatbots as a companion rather than social ro-

bots, unless the physical presence is a critical element.

4 TOWARDS AN AUGMENTED

TASK SHARING

In summary, smart machines may have three compar-

ative advantages compared to teachers.

First, smart machines can handle complexity very

well due to their ability to collect, curate, process and

analyse large amounts of data. No matter how many

students or simultaneous inputs, the smart machine

does not forget and can serve the teacher by providing

analytical decision support. In addition, smart ma-

chines are, due to their nature, ideally suited to regu-

late, control, and manage the digital classroom eco-

system on behalf of the teacher.

Second, smart machines are good at tasks with a

low variety, because it is easier to train smart ma-

chines on tasks which are repetitive. In the classroom,

such repetitive tasks could be for example the correc-

tion of assignments or exams. For human teachers

those tasks are often boring and they may make mis-

takes over time. A smart machine does not get bored

and can correct all exam questions which are not char-

acterized by uncertainty or equivocality.

Third, smart machines have a high attractivity to

take over selected classroom tasks, because the wage

costs of human teachers in industrialized countries are

high compared to the operating costs of smart ma-

chines. In particular, chatbots as representatives of

smart machines are attractive, as they are cheaper than

social

robots due to their lack of a physical presence.

To put it another way: For a given budget, the quality

of teaching can be improved by realising comparative

advantages.

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing

Source: Based on Davenport & Kirby (2016); Daud et. al. (2017) and own contributions.

Figure 4: Augmentation strategies of teachers in relation to smart machines.

Against this backdrop, the question is: How

should teachers position themselves in relation to

smart machines to be able to engage in an augmented,

mutually reinforcing way of task sharing?

According to Davenport and Kirby (2016), five

augmentation strategies are possible, which are perti-

nent to different occupational groups, but especially

to knowledge workers. Table 4 shows an adaptation

of the five augmentation strategies to the teaching

profession and indicates how teachers could position

themselves in relation to smart machines; teachers

have various options to engage with smart machines

in an augmented task sharing.

According to the Step In strategy, teachers could

train the smart machine and shift tasks to it. This cre-

ates added value because the smart machine can re-

lieve the teacher of work. The human teacher is

needed for this training process as it involves to some

degree uncertainty and equivocality. Only the teacher

can decide on the appropriate training tasks and

measures to be applied.

The Step Up strategy is similar to the concept of

Dillenbourg (2013), who introduced the concept of

“orchestration” of learning activities as real time

management for distributed activities over the class-

room ecosystem. In the Step Up strategy, the teacher

could concentrate on higher level tasks inside the

classroom. Similar to a conductor of a concert (Shah-

moradi et al., 2020), the teacher orchestrates and man-

ages the classroom and its players. He or she keeps an

overview, evaluates and decides on the ethical use of

a smart machine. As these tasks involve a high degree

of uncertainty, complexity and equivocality, a human

teacher is needed. The smart machine can support the

teacher in this process, by serving as an interface to

the functions of the classroom ecosystem. The smart

machine further amplifies the cognitive strengths of

the human teacher by making recommendations and

providing decision support.

Human teachers could also Step Forward and par-

ticipate in the content development and the data anal-

ysis of the smart machine. They could control for data

biases of the smart machine and share content with

other teachers. Through this, they could contribute to

a long-term improvement of the smart machine and

its applications. In this process, positive and negative

aspects have to be weighed against each other, which

is why the process is characterized by uncertainty and

equivocality.

According to the Step Aside strategy, teachers

could take on tasks that go beyond information pro-

cessing (complexity) or require tacit knowledge.

Teachers could increasingly take on the role of a

coach, who communicates the learning content pro-

vided by the smart machine in a didactically appeal-

ing way and assists the learners in an advisory role.

The teacher is supported in this process by the smart

machine, for example through the means of learning

analytics. From the smart machine, students could

also receive additional prompts to plan their own

learning processes more effectively and improve their

metacognitive learning strategies (Bräuer, 2003). The

Smart machine

(digital, autonomous and intelligent)

Regulation, control and management

of the Classroom ecosystem on behalf

of the teacher

To show discovered knowledge,

to check and correct for data biases

To show recommendations

Step Aside / Narrowly

To use, interact, participate, and communicate

Students

Step Forward / In

To design, develop and train

Teachers as Content-Developers

Teachers as Data Scientists

To give recommendations

To evaluate recommendations

Teachers as Coaches

Teachers as Managers

To orchestrate the classroom and its players,

to decide on the ethical use of smart machines

Step Up

Uncertainty, Complexity, Equivocality

Uncertainty, Equivocality

Complexity, Variety, Attractivity

Equivocality, Preferences, Variety

Classroom Ecosystem

(Content preparation, -presentation and

distribution, interaction and engagement,

automated evaluation)

CSEDU 2021 - 13th International Conference on Computer Supported Education

Step Aside approach is characterized by equivocality

(e.g., discussing), but also by preferences (e.g., moti-

vating), which is why a human teacher is needed.

Finally, in the Step Narrowly strategy, the teacher

could perform tasks that cannot be performed well by

smart machines (variety). This could include non-re-

petitive tasks as the individual coaching of students

with different needs or the maintenance of the smart

machine.

Figure 4 summarizes the conceptual framework

with the different augmentation strategies of teachers

in relation to smart machines. It is important to point

out, that teachers can follow multiple strategies and

do not have to choose just one. For example, during

the lecture, teachers could use the Step Up and Step

Aside strategy and switch between their roles as man-

agers and coaches.

With our framework, we provide a guideline for

an augmented task sharing based on the relative

strengths of teachers and smart machines. We high-

light ways how teachers could collaborate with smart

machines, and how they may leverage their capabili-

ties through smart machines.

5 SUMMARY AND OUTLOOK

Our conceptual framework is based on the theory of

comparative advantage. Drawing from this, we have

shown ways how an augmented task sharing based on

relative strengths of teachers and smart machines

could look like.

How the task sharing in the classroom will look

like in the future is ultimately an empirical question;

relative strengths heavily depend on student percep-

tions that could be empirically investigated. With our

framework, we want to contribute to a better under-

standing of the concepts and variables that should be

considered when investigating task sharing of teach-

ers and smart machines. In a next step, empirical re-

search could further investigate the relative strength

dimensions to get a better understanding of which

tasks could be assigned to smart machines.

AI has currently triggered a second wave of digi-

talization, in which data is not only stored and pro-

cessed digitally (first wave) but also automatically in-

terpreted and actively used by intelligent algorithms

(Wahlster, 2017). While schools and teachers are still

absorbing and integrating the first wave of digitaliza-

tion (Whalster, 2017) into their curriculum, another

wave of digitalization is already rolling in. Due to the

novelty and complexity of the topic, there is a risk that

teachers will be overwhelmed by smart machines and

will not know how to use them in teaching. To pre-

vent this, prospective teachers should be equipped

with the necessary knowledge, skills, and attitudes to

see the opportunities in the use of smart machines ra-

ther than dangers. Teachers should be enabled and

supported to sit in the driver seat for shaping their

school in the current major transition phase.

AI transformation does not mean that less teachers

are needed (Dillenbourg, 2016). However, the role of

the teacher may change. Just as paper, white boards

or computer tablets – smart machines have the poten-

tial to become part of the educational infrastructure,

delivering a learning experience tailored to the learner

and relieving the burden on teachers where necessary

(Belpaeme et al, 2018, p. 7). Such individual support

could particularly be beneficial for disadvantaged

learners. Currently, the use of smart machines in ed-

ucational institutions may be limited due to technical

and logistical challenges (Belpaeme et al., 2018, p. 7),

but as technology becomes cheaper and better, the use

of smart machines in education is likely to increase.

Through our conceptual framework, we aim at a

better understanding of the digital transformation

from a teacher perspective. However, as many pre-

service and in-service teachers are not ready to sup-

port and adopt new technologies related to AI, effec-

tive teacher education and continuing education pro-

grams have to be designed and offered to support the

adoption of these new technologies (Luan et. al.,

2020, p. 7). There is a need for more robot-proof skills

and strategies, that make it possible to cooperate suc-

cessfully with smart machines without being replaced

by them in the long term (Aoun, 2017).

REFERENCES

Aoun, J. E. (2017). Robot-proof: higher education in the

age of artificial intelligence. MIT press.

Abildgaard, J.R., Scharfe, H. (2012). A Geminoid as Lec-

turer. In: Ge, S.S., Khatib, O., Cabibihan, JJ., Simmons

R., Williams, MA. (Eds.) Social Robotics. ICSR 2012.

Lecture Notes in Computer Science: Vol 7621.

Springer, Berlin, Heidelberg.

https://doi.org/10.1007/978-3-642-34103-8_41

Baldwin, R., & Forslid, R. (2020). Globotics and develop-

ment: When manufacturing is jobless and services are

tradable (Working paper 26731). National Bureau of

Economic Research. https://doi.org/10.3386/w26731

Belpaeme, T., Kennedy, J., Ramachandran, A., Scassellati,

B., & Tanaka, F. (2018). Social robots for education: A

review. Science robotics, 3(21), 1-9.

https://doi.org/10.1126/scirobotics.aat5954

Bolea Monte, Y., Grau Saldes, A., & Sanfeliu Cortés, A.

(2016). From research to teaching: Integrating social

robotics in engineering degrees. International Journal

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing

of Computer, Electrical, Automation, Control and In-

formation Engineering, 10(6), 1020-1023.

https://doi.org/10.5281/zenodo.1124667

Braga, C. P., Drake-Brockman, J., Hoekman, B., Bradford

Jensen, J., Low, P., Momdouh, H., Sauvé, P.,

Schwarzer, J., & Stephenson, S. (2019). Services Trade

for Sustainable, Balanced, and Inclusive Growth.

https://t20japan.org/policy-brief-services-trade-sus-

tainable-inclusive-growth/

Bräuer, G. (2003). Schreiben als reflexive Praxis:

Tagebuch, Arbeitsjournal, Portfolio. Fillibach.

Brynjolfsson, E., & McAfee, A. (2014). The second ma-

chine age: Work, progress, and prosperity in a time of

brilliant technologies. New York: Norton.

Byrne, B. M., Rossi, A., & Doolan, M. (2017). Humanoid

Robots Attending Lectures: Research Informed Teach-

ing. In Proceedings of the 2017 9th International Con-

ference on Education Technology and Computers, 7-

11. https://doi.org/10.1145/3175536.3175556

Chua, X.N. & Chew. E. (2015). The Next Wave of Learning

with Humanoid Robot: Learning Innovation Design

starts with “Hello NAO”. In T. Reiners, B.R. von Kon-

sky, D. Gibson, V. Chang, L. Irving, & K. Clarke

(Eds.), Globally connected, digitally enabled. Proceed-

ings Ascilite 2015, Perth, 52-56. http://hdl.han-

dle.net/10369/9531

Cooney, M., & Leister, W. (2019). Using the engagement

profile to design an engaging robotic teaching assistant

for students. Robotics, 8(1), 21-47.

https://doi.org/10.3390/robotics8010021

Daud, A., Aljohani, N. R., Abbasi, R. A., Lytras, M. D.,

Abbas, F., & Alowibdi, J. S. (2017). Predicting student

performance using advanced learning analytics. In Pro-

ceedings of the 26th international conference on World

Wide Web companion, 415-421.

https://doi.org/10.1145/3041021.3054164

Davenport, T. H., & Kirby, J. (2016). Only humans need

apply: Winners and losers in the age of smart machines.

New York, NY: Harper Business.

Dillenbourg, P., Nussbaum, M., Dimitriadis, Y., &

Roschelle, J. (2013). Design for classroom orchestra-

tion. Comput. Educ., 69, 485-492.

https://doi.org/10.1016/j.compedu.2013.04.013

Dillenbourg, P. (2016). The evolution of research on digital

education. International Journal of Artificial Intelli-

gence in Education, 26(2), 544-560.

https://doi.org/10.1007/s40593-016-0106-z

Floridi, L. (2013). The ethics of information. Oxford Uni-

versity Press. www.doi.org/10.1093/ac-

prof:oso/9780199641321.001.0001

Floridi, L. (2016). Hyperhistory, the emergence of the

MASs, and the design of infraethics. Information, Free-

dom and Property: The Philosophy of Law Meets the

Philosophy of Technology, 153.

Flynn, P. (2017). Using an Autonomous Humanoid Robot

as a Pedagogical Platform in the Business Classroom.

Journal of Social Science Studies, 4(1), 178-188.

https://doi.org/10.5296/jsss.v4i1.10227

Gao, Y., Barendregt, W., Obaid, M., & Castellano, G.

(2018). When robot personalisation does not help: In-

sights from a robot-supported learning study. In 2018

27th International Symposium on Robot and Human In-

teractive Communication, 705-712. IEEE.

https://doi.org/10.1109/ROMAN.2018.8525832

Guggemos, J., Seufert, S., & Sonderegger, S. (2020). Hu-

manoid robots in higher education: Evaluating the ac-

ceptance of Pepper in the context of an academic writ-

ing course using the UTAUT. British Journal of Edu-

cational Technology, 51(5), 1864-1883.

https://doi.org/10.1111/bjet.13006

Harteis, C., & Billett, S. (2013). Intuitive expertise: Theo-

ries and empirical evidence. Educational Research Re-

view, 9, 145-157.

https://doi.org/10.1016/j.edurev.2013.02.001

Jarrahi, M. H. (2018). Artificial intelligence and the future

of work: Human-AI symbiosis in organizational deci-

sion making. Business Horizons, 61(4), 577-586.

https://doi.org/10.1016/j.bushor.2018.03.007

Landsburg, L. F., (n.d.). Comparative Advantage. The Li-

brary of Economics and Liberty.

https://www.econlib.org/library/Topics/Details/com-

parativeadvantage.html

Latham, S., & Humberd, B. (2018). Four ways jobs will re-

spond to automation. MIT Sloan Management Review,

60(1), 11-14.

Lehmann, H., & Rossi, P. G. (2019). Social robots in edu-

cational contexts: Developing an application in enac-

tive didactics. Journal of e-Learning and knowledge

Society, 15(2). https://doi.org/10.20368/1971-

8829/1633

Luan, H., Geczy, P., Lai, H., Gobert, J., Yang, S. J., Ogata,

H., Baltes, J., Guerra, R., Li, P., & Tsai, C. (2020).

Challenges and future directions of big data and artifi-

cial intelligence in education. Frontiers in psychology,

1-11. http://dx.doi.org/10.3389%2Ffpsyg.2020.

580820

Macfadyen, L. P. (2017). Overcoming barriers to educa-

tional analytics: How systems thinking and pragmatism

can help. Educational Technology, 31-39.

https://www.jstor.org/stable/44430538

Makridakis, S. (2017). The forthcoming Artificial Intelli-

gence (AI) revolution: Its impact on society and firms.

Futures, 90, 46-60. https://doi.org/10.1016/j.futu-

res.2017.03.006

Masuta, H., Kawamoto, T., Sawai, K., Motoyoshi, T., Ta-

mamoto, T., Koyanagi, K. I., & Oshima, T. (2018).

Presentation Robot System with Interaction for Class.

In 2018 Symposium Series on Computational Intelli-

gence (SSCI), 1801-1806. IEEE.

https://doi.org/10.1109/SSCI.2018.8628804

Pereira, A. (2019, March 15). What are smart machines?

Career in STEM. https://careerinstem.com/what-are-

smart-machines/

Reich-Stiebert, N., Eyssel, F., & Hohnemann, C. (2019).

Exploring university students’ preferences for educa-

tional robot design by means of a user-centered design

approach.

International Journal of Social Robotics, 12,

1-11. https://doi.org/10.1007/s12369-019-00554-7

CSEDU 2021 - 13th International Conference on Computer Supported Education

Ricardo, D. (1891). Principles of political economy and

taxation. G. Bell and sons.

Ruffin, R. (2002). David Ricardo's discovery of compara-

tive advantage. History of political economy, 34(4),

727-748.

Saini, M. K., & Goel, N. (2019). How smart are smart class-

rooms? A review of smart classroom technologies.

ACM Computing Surveys (CSUR), 52(6), 1-28.

https://doi.org/10.1145/3365757

Shahmoradi, S., Kothiyal, A., Olsen, J. K., Bruno, B., &

Dillenbourg, P. (2020, September). What Teachers

Need for Orchestrating Robotic Classrooms. In Euro-

pean Conference on Technology Enhanced Learning,

87-101. Springer, Cham. https://doi.org/10.1007/978-

3-030-57717-9_7

Sharkey, A. J. (2016). Should we welcome robot teachers?.

Ethics and Information Technology, 18(4), 283-297.

https://doi.org/10.1007/s10676-016-9387-z

Thimm, C., Regier, P., Cheng, I. C., Jo, A., Lippemeier, M.,

Rutkosky, K., Bennewitz, M., & Nehls, P. (2019). Die

Maschine als Partner? Verbale und non-verbale

Kommunikation mit einem humanoiden Roboter. In

Die Maschine: Freund oder Feind?, 109-134. Springer

VS, Wiesbaden. https://doi.org/10.1007/978-3-658-

22954-2_6

van den Berghe, R., Verhagen, J., Oudgenoeg-Paz, O., Van

der Ven, S., & Leseman, P. (2019). Social robots for

language learning: A review. Review of Educational

Research, 89(2), 259-295. https://doi.org/10.3102/

0034654318821286

Wahlster, W. (2017, June). Künstliche Intelligenz als

Treiber der zweiten Digitalisierungswelle. IM+io Das

Magazin für Innovation, Organisation und

Management, 2, 10-13.

Wilson, H. J., & Daugherty, P. R. (2018). Collaborative in-

telligence: humans and AI are joining forces. Harvard

Business Review, 96(4), 114-123.

Relative Strengths of Teachers and Smart Machines: Towards an Augmented Task Sharing