Of ‘black boxes’ and algorithmic decision-making in (higher) education

Abstract

Higher education institutions have access to higher volumes and a greater variety and granularity of student data, often in real-time, than ever before. As such, the collection, analysis and use of student data are increasingly crucial in operational and strategic planning, and in delivering appropriate and effective learning experiences to students. Student data – not only in what data is (not) collected, but also how the data is framed and used – has material and discursive effects, both permanent and fleeting. We have to critically engage claims that artificial intelligence and the ever expansive/expanding systems of algorithmic decision-making provide speedy, accessible, revealing, panoramic, prophetic and smart analyses of students' risks, potential and learning needs. We need to pry open the black boxes higher education institutions (and increasingly venture capital and learning management system providers) use to admit, steer, predict and prescribe students’ learning journeys.

Keywords

Algorithmic-decision making higher education learning analytics student data

This article is a part of special theme on The Black Box Society. To see a full list of all articles in this special theme, please click here: https://journals.sagepub.com/page/bds/collections/revisitingtheblackboxsociety

Introduction

In this brief commentary, I map the social imaginary (and its impact) pertaining to algorithmic decision-making in the context of the measurement, collection, analysis and use of student data, a practice known as learning analytics, in higher education. This imaginary allows ‘data themselves to come to life and begin to have consequences when they are analysed and when those analyses are integrated into social, governmental and organisational structures’ (Beer, 2019: 14). The data imaginary and gaze include not only those who have the power to define, collect, analyse and use data, but also those whose data is collected, analysed and used. As such, this ‘data gaze’ in both ‘its material and discursive configurations’ (Beer, 2019: 7) affects us all – whether as the one who gazes at or the one who is gazed upon. In higher education in particular, this data gaze functions as a ‘one-way mirror’ (Pasquale, 2015: 9) and students (and increasingly faculty) have no way to escape this gaze, or at best, return the gaze. The collection, analyses and use of student data furthermore resembles an ‘algocracy’ described as the normalisation and naturalisation of governance by algorithms that exclude, per se, any disagreement or user interrogation, or other discursive or material options (Aneesh, 2006, 2009; see also Danaher, 2016a, 2016b). For students, there is ‘no exit’ (Pasquale, 2015: 52) and surveillance-as-service is an integral part of terms and conditions (Khalil et al., 2018).

Higher education can be classified as a ‘data frontier’ (Beer, 2019: 19), as a data-unchartered territory where data-led decision-making is still in its relative infancy, and where ‘those invested in this industry are inevitably seeking to push these boundaries back’ (Beer, 2019: 19). The notion of the ‘frontier’ is fraught with historical narratives of conquest, colonisation and subjugation (Gouge, 2007; Kwet, 2018), and, as such, adds a specific heuristic lens to understanding the datafication of learning. As ‘the frontier’ institutions have access to higher volumes and a greater variety and granularity of student data, often in real-time, than ever before (Slade et al., 2019). As such, higher education is part of the ‘data revolution’ and ‘data deluge’ (Kitchin, 2014) with many believing that more or big(ger) data is necessarily better data (Prinsloo et al., 2015). Data, whether small or big, is like technology ‘neither good nor bad; nor is it neutral … technology’s interaction with the social ecology is such that technical developments frequently have environmental, social, and human consequences that go far beyond the immediate purposes of the technical devices and practices themselves' (Kranzberg, 1986: 545). As Pasquale (2015) states: ‘ … we should never lose sight of the fact that the numbers on their computer terminals have real effects, deciding who gets funded and found, and who is left discredited or obscure’ (215; see also Eynon, 2013).

A growing part of the data imaginary of higher education is ‘authority … increasingly expressed algorithmically’ (Pasquale, 2015: 8), with the power ‘to include, exclude, and rank to ensure that certain public impressions become permanent, while others remain fleeting’ (Pasqual, 2015: 14; see also Prinsloo, 2017, 2019). Data and data scientists in particular ‘must recognize themselves as political actors engaged in normative constructions of society and, as befits political work, evaluate their work according to its downstream material impacts on people’s lives’ (Green, 2018: 1).

The lure of faster, smarter and more effective

Amid the various issues and often dramatic changes facing higher education (e.g., Altbach et al. (2009), higher education increasingly finds its security in the mantra of evidence-based decision-making (Gagliardi et al., 2018; Prinsloo, 2016), ‘performativities and fabrications’ (Ball, 2004: 143), managerialism (Lynch et al., 2015) and new public management (Bleiklie, 2018). The culture of performativity is a ‘technology, a culture and a mode of regulation, or even a system of “terror” … , that employs judgements, comparisons, and displays of means of control, attrition, and change’ (Ball, 2004: 144). These quantifications and metrics

facilitate the making and remaking of judgements about us, the judgements we make of ourselves and the consequences of those judgements as they are felt and experienced in our lives. We play with metrics and we are more often played by them. (Beer, 2016: 3)

What we forget in our obsession with numbers and quantification is that when ‘humans are summed, aggregated and accounted for; then much remains forgotten, unsaid, concealed’ (Eldin, 2006, in Beer, 2016: 59–60). In the words of Pasquale (2015), the data imaginary creates ‘a rule of scores and bets’ (191), it ‘creates invisible powers’ (193) and increases unfair competition.

Algorithmic decision-making in higher education functions as a ‘prosthetic vision’ (Beer, 2019: 7) allowing automated insights into data that were not only not possible before, but also not accessible for those without the necessary training and knowledge. As such the data gaze presents itself as ‘speedy, accessible, revealing, panoramic, prophetic and smart’ (Beer, 2019: 22; emphasis in the original). In order to realise these six characteristics, the human element in relation to and involvement in algorithmic decision-making systems is increasingly taken out of the equation (Knox, 2010: 211; Prinsloo, 2017, 2019). Danaher (2015) developed a tentative model to classify algocratic decision-procedures using the four tasks of sensing, processing, acting and learning as basis for mapping. Algocratic modes of decision-making or algocratic systems refer to operational or governance systems where algorithms act or ‘rule’ with or without a range of human supervision or engagement – from where human actors are supported by algorithms, to humans sharing tasks with algorithms, to algorithms performing tasks with human supervision, to automated independent algorithms (Danaher, 2015). Realising the characteristics of ‘speedy, accessible, revealing, panoramic, prophetic and smart’ however inherently points to the exclusion of human participation in the ‘black box’.

The ‘black box’ of algorithmic decision-making or algocracy delivers speedy, real-time information based on real-time analysis, on a continuous basis, closing the ‘gap between data and knowledge’ (Beer, 2019: 23). As such it makes us ‘addicted to speed’ (Pasquale, 2015: 195). Software provides the analysis and as such, the analysis is accessible to anyone – even those with no experience, analytical or technical expertise. Circumventing the subjectivity of human judgement, algorithmic decision-making reveals that which is hidden to the naked eye and as such acts as séance (Prinsloo, 2016) and panoramic, acting as ‘prosthetic eye’, all-seeing and omnipotent (Beer, 2019: 25). The ‘black box’ also has the ability to ‘grasp the future and use it in the present’ (Beer, 2019: 27) and fulfils a prophetic role, opening up new possibilities. As such algorithmic decision-making systems produce and enable futures and outcomes in smart and responsive ways. Self-learning algorithms increasingly also act autonomously, without human involvement./supervision. Actually, to fully realise ‘speedy, accessible, revealing, panoramic, prophetic and smart’ (Beer, 2019: 22), humans and human oversight and supervision should be removed from the equation. These six characteristics also require data to be increasingly flattened, faster than before and individuals fitted into historical categories of race, household, income and marital status while there is increasing evidence that these categories may, in fact, be part of a ‘zombie sociology’ (Beck, 2001) and act as ‘zombie categories’ (Gullion, 2018: 68). For example, when we assume ‘household’ as category, what do we mean, include and exclude?

So where does this leave the collection, analysis and use of student data by algorithmic decision-making systems?

Student data and algorithmic decision-making

When ‘The black box society’ (Pasquale, 2015) was published in 2015, learning analytics as research field and practice was already four years old. At the inaugural learning analytics conference in 2011 in Banff, Canada, the Conference Call made mention of the reality that the ‘growth of data surpasses the ability of organizations to make sense of it’ (Siemens, 2010). The Conference Call also states that ‘[l]earning institutions and corporations make little use of the data learners “throw off” [sic] in the process of accessing learning materials, interacting with educators and peers, and creating new content’ (Siemens, 2010). Among the reasons necessitating a more intentional and integrated collection, analysis and use of student data, the organisers mention, inter alia, pressures ‘to reduce costs and increase efficiency’ as well as ‘increased competitiveness and productivity’ (ibid.). While algorithmic decision-making, artificial intelligence (AI), machine learning and neural networks are not mentioned per se, the organisers do mention ‘[a]dvances in knowledge modelling and representation, the semantic web, data mining, analytics, and open data form a foundation for new models of knowledge development and analysis’ (ibid.). The Conference Call concludes with proposing the first seminal definition of learning analytics as ‘the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs’ (ibid.).

In the same year of the learning analytics inaugural conference (LAK’11), the number of scholarly articles mentioning ‘learning analytics’ in combination with ‘artificial intelligence’ equals 46 results (excluding patents and citations). For example, Elias (2011) mentions that the ‘University of Phoenix and Capella University consistently make extensive use of artificial intelligence and predictive modelling in marketing, recruitment, and retention and have shaped their cultures around performance’ (16). In referring to the tool set available to learning analytics, Elias mentions ‘data visualization, decision trees, neural networks, regression analysis, machine learning, and artificial intelligence’ (12). Also in 2011, Blikstein (2011) mentions two examples where AI was used, one in a computer-based application where a machine learning algorithm allowed them to ‘to most accurately classify each student in terms of their content knowledge, based on comparisons with expert-formulated responses’ (110). The other example refers to identifying student sentiment through studying and coding students’ facial expressions and applying machine learning to the data produced. Of specific interest is the value added by AI techniques and the following are mentioned: ‘1) detailed, multimodal student activity data (gestures, sketches, actions) as a primary component of analysis, 2) automation of data capture and analysis, 3) multidimensional data collection and analysis’ (111). In broad terms, these may fall in the category of surveillance and analysis as discussed above. Outside of algorithmic decision-making as surveillance, there is also evidence of algorithmic decision-making acting autonomously. For example, Verbert et al. (2011) report on small-scale experiments with recommender systems and state that while

such experiments offer valuable insights into the usefulness and relevancy of recommender systems for learning, stronger conclusions about the validity and generalizability of scientific experiments could be drawn if researchers have the possibility of verification, repeatability, and comparisons of results based on large datasets that capture learner interactions in real setting. (44)

Fast-forward to the year 2018 where the total number of scholarly articles mentioning ‘learning analytics’ in combination with ‘artificial intelligence’ numbers 1670. So far, in 2019, the same search terms (on 28 May 2019) show 644 results. It is clear that algorithmic decision-making systems have become an integral part of the broader research agenda at the intersections of learning, media and technology. For example, Williamson et al. (2019) puts the increasing collection, analysis and use of student data central on the agenda and ask: ‘ … what problems should education researchers be identifying and pursuing that can help flesh out wider understandings of the new political of the digitized and datafied twenty-first century?’ (88). Key to the extraction and use of student data is the scope and impact of the use of machine learning and AI as new forms of ‘nonconscious cognition’ (Hayles, 2017) that create a ‘temporal gap between human and technical cognition’(142) and act, with other actors, in a ‘complex ecology that [ … ] can be understood as swarm behaviour’ (163). Evidence of the increasing scope of the data gaze and swarm behaviour in education is provided in the surveillance of students outside the original purpose of increasing the effectiveness of learning spilling over into a ‘One view. Total awareness’ offered by surveillance systems such as Social Sentinel (Herold, 2019).

Avoiding Frankenstein and Kafka

While there is evidence of the huge potential of algorithmic decision-making systems to solve problems or to contribute to human flourishing, there are equally very real concerns about systems. Consider for a moment using AI to identify and inform cancer treatment (Coccia, 2020), forecasting the progression of the COVID-19 in China (Hu et al., 2020) and concerns with regarding the roll out of AI-driven surveillance systems to identify and control individuals’ movement and risk (Kupferschmidt and Cohen, 2020), and the persistence of racial bias in AI systems (Allen, 2019). Amid the many concerns and uncertainties about the various aspects of algorithmic decision-making systems, likening these algorithmic decision-making systems to the monster created by Dr Frankenstein comes effortlessly as we have to recognise AI and algorithmic decision-making systems as the result of human endeavour, and as such, these systems are our creation (Latour, 2012; Prinsloo, 2017). We have to face our creature, and increasingly our creature eludes our gaze and hides in ‘black boxes’, and worse still, turns their gaze upon us. Algorithms and algorithmic decision-making systems are ‘not just abstract computational processes; they also have the power to enact material realities by shaping social life to various degrees’ (Butcher, 2017: 40) and individuals have, increasingly, very little control. I have a sense that we are condemned to an ‘ouroborotic Kafkaesque journey where the tensions between the dangers and the potentials of algorithmic decision-making are never (really) resolved’ (Prinsloo, 2017: 144). These tensions are furthermore visible in research on using algorithmic decision-making systems to determine who gets accepted into institutions, who has access to student funding and support, and which students will not be ‘worth’ the investment of additional support to retain them in a program (Johnson, 2017; Jones and McCoy, 2019; Prinsloo and Slade, 2014, 2017).

In the specific context of students’ engagement with an institution in the context of their learning, we have to critically consider the scope and function of automation, how visible is the algorithm and its assumptions and functions, how engaged is the data-object with the collection and analysis of data, the scope and function of these data assemblages, the temporal frame of collection, analysis and use, etc. (see Knox, 2010; Prinsloo, 2017). This aligns with Pasquale’s (2015) suggestion of ‘watching (and improving) the watchers’ (140). We need to pry open the black boxes higher education institutions (and increasingly venture capital and learning management system providers) use to admit, steer, predict and prescribe students’ learning journeys. Students have a right not only of access, but also to provide additional information and correct institutions’ assumptions and categorisations of them (Slade and Prinsloo, 2013).

There is evidence in the learning analytics community of scholars, researchers and practitioners, of an increasing recognition of and sensitivity towards the many issues in the context of algorithmic decision-making in learning analytics. For example, Shum (2017) acknowledges that learning analytics functions as a ‘black box’ and he proposes an accountability analysis that focuses on system integrity from various perspectives and that pushes thinking beyond many of the current thinking about the ethical issues in learning analytics. Chen and Zhu (2019) propose Value Sensitive Design as a ‘systemic approach with specific strategies and methods to help researchers and designers explicitly incorporate the consideration of human values into design’ (344). Their research foreground the tension between utility and the ‘values of students’ autonomy, privacy, social well-being and self-image’ (347; emphasis in the original). Other values include freedom from bias, fairness, epistemic agency and ease of information seeking (Chen and Zhu, 2019).

To what extent these steps will make a difference in the context of the commercial value embedded in the ‘data gaze’ (Beer, 2019) and ‘black boxes’ in service of the ‘lords of the information age’ entrenching ‘a digital aristocracy’ (Pasquale, 2015: 218) will have to be seen. Up to then, we are called to face and dialogue with and care for our monster (Latour, 2012).

Conclusion

Despite a greater sensitivity to the ethical issues surrounding the ‘black box’ of collection, analysis and use of student data (Slade and Prinsloo, 2013), and practice-oriented proposals for algorithmic accountability (Shum, 2017), and values (Chen and Zhu, 2019), these proposals, frameworks and methods may not have resolved ‘the normative questions of what impacts are desirable and how to negotiate between conflicting perspectives (nor the practical question of how to leverage technology toward these ends)’ (Green, 2018: 6). And though codes of ethics, accountability and transparency may assist in addressing the issues arising from the ‘black box’ of the collection, analysis and use of student data, central to (re)considering Pasquale’s (2015) ‘The black box society’ is the recognition of the inherently political nature of data. Data is not raw and/or objective but is entangled in ‘specific arrangements and interests in the nexus of economic, political, social, technological, environmental and legal apparatuses, structures, and elements’ and realise, increasingly autonomously ‘ideological and political ontologies and epistemologies’ (Prinsloo, 2019: 3).

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Paul Prinsloo

References

Allen

(2019) The colour of algorithms: An analysis and proposed research agenda for deterring algorithmic redlining. Fordham Urban Law Journal 46: 219–270.

Altbach

Reisberg

Rumbley

(2009) Trends in global higher education: Tracking an academic revolution. Available at: https://www.sensepublishers.com/media/1207-trends-in-global-higher-education.pdf (accessed 2 September 2019).

Aneesh

(2006) Virtual Migration: The Programming of Globalization. Durham, NC: Duke University Press.

Aneesh

(2009) Global labour: Algocratic modes of organization. Sociological Theory 27(4): 347–370.

Ball

(2004) Performativities and fabrications in the education economy: Towards the performative society. In: Ball

(ed) The Routledge Falmer Reader in Sociology of Education. London: RoutledgeFalmer, pp.143–155.

Beck

(2001) Interview with Ulrich Beck. Journal of Consumer Culture 1(2): 261–277.

Beer

(2016) Metric Power. London: Palgrave.

Beer

(2019) The Data Gaze. London: Sage.

Bleiklie, I. (2018). New public management or neoliberalism, higher education. In: Shin JC and Teixeira P (eds.) Encyclopedia of International Higher Education Systems and Institutions, pp.1–6. Springer Science+Business Media Dordrecht.

10.

Blikstein

(2011) Using learning analytics to assess students’ behaviour in open-ended programming tasks. In: Proceedings of the 1st international conference on learning analytics and knowledge, February, pp.110–116. New York: ACM.

11.

Chen

Zhu

(2019) Towards value-sensitive learning analytics design. In: Proceedings of the 9th international conference on learning analytics & knowledge, March, pp.343–352. New York: ACM.

12.

Coccia

(2020) Deep learning technology for improving cancer care in society: New directions in cancer imaging driven by artificial intelligence. Technology in Society 60: 101198.

13.

Danaher

(2015) How might algorithms rule our lives? Mapping the logical space of algocracy. Available at: https://philosophicaldisquisitions.blogspot.com/2015/06/how-might-algorithms-rule-our-lives.html (accessed 2 September 2019).

14.

Danaher

(2016a) The threat of algocracy: Reality, resistance and accommodation. Philosophy and Technology 29(3): 1–24.

15.

Danaher

(2016b) The logical space of algocracy (Redux). Available at: http://philosophicaldisquisitions.blogspot.co.za/2016/11/the-logical-space-of-algocracy-redux.html (accessed 2 September 2019).

16.

Elias

(2011) Learning analytics. Learning 1–22.

17.

Eynon R (2013) The rise of big data: What does it mean for education, technology, and media research? Learning, Media and Technology 38(3): 237–240. DOI: 10.1080/17439884.2013.771783

18.

Gagliardi

Parnell

Carpenter-Hubin

(2018) The analytics revolution in higher education. Change: The Magazine of Higher Learning 50(2): 22–29.

19.

Gouge C (2007) The American frontier: History, rhetoric, concept. Americana: The Journal of American Popular Culture (1900-Present) 6(1). Available at: https://search.proquest.com/docview/1519971854?accountid=14648 (accessed 6 June 2020).

20.

Green

(2018) Data science as political action: Grounding data science in a politics of justice. arXiv preprint arXiv:1811.03435.

21.

Gullion

(2018) Diffractive Ethnography: Social Sciences and the Ontological Turn. New York: Routledge.

22.

Hayles

(2017) Unthought: The Power of the Cognitive Nonconscious. Chicago, IL: University of Chicago Press.

23.

Herold

(2019) Schools are deploying massive digital surveillance systems. The results are alarming. Education Week, 30 May. Available at: https://www.edweek.org/ew/articles/2019/05/30/schools-are-deploying-massive-digital-surveillance-systems.html (accessed 2 September 2019).

24.

Jin

, et al. (2020) Artificial intelligence forecasting of covid-19 in china. arXiv preprint arXiv:2002.07112.

25.

Johnson JA (2017) Structural justice in student analytics, or, the silence of the bunnies. In: Paper presented at the annual meeting for the Eastern Sociological Society: 2017 digital sociology mini-conference, Philadelphia, 24 February. Available at: https://the-other-jeff.com/wpcontent/uploads/2017/02/Structural-Justice-in-Learning-Analytics-ESS.pdf (accessed 6 June 2020).

26.

Jones

McCoy

(2019) Reconsidering data in learning analytics: Opportunities for critical research using a documentation studies framework. Learning, Media and Technology 44(1): 52–63.

27.

Khalil M, Prinsloo P and Slade S (2018) User consent in MOOCs – Micro, meso, and macro perspectives. The International Review of Research in Open and Distributed Learning 19(5). Available at: https://doi.org/10.19173/irrodl.v19i5.3908 (accessed 6 June 2020).

28.

Kitchin

(2014) The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences. London: Sage.

29.

Knox

(2010) Spies in the house of learning: A typology of surveillance in online learning environments. In: Proceedings from e-Learning: The Horizon and beyond conference, St John’s, Newfoundland, Canada.

30.

Kranzberg

(1986) Technology and history: Kranzberg’s laws. Technology and Culture 27(3): 544–560.

31.

Kupferschmidt K and Cohen J (2020) China’s aggressive measures have slowed the coronavirus. They may not work in other countries. Science, 2 March. Available at: https://www.sciencemag.org/news/2020/03/china-s-aggressive-measures-have-slowed-coronavirus-they-may-notwork-other-countries. (accessed 6 June 2020).

32.

Kwet M (2018) Digital colonialism: US Empire and the new imperialism in the global South. Race & Class 60(4): 3–26. DOI: 10.1177/0306396818823172.

33.

Latour

(2012) Love your monsters. Why we must care for our technologies as we do our children. Available at: http://thebreakthrough.org/index.php/journal/past-issues/issue-2/love-your-monsters/ (accessed 2 September 2019).

34.

Lynch

Grummell

Devine

(2015) New Managerialism in Education: Commercialisation, Carelessness and Gender. London: Palgrave Macmillan.

35.

Pasquale F (2015) The black box society. Cambridge, MA: Harvard University Press.

36.

Prinsloo

(2016) Evidence-based decision making as séance: Implications for learning and student support. In: Botha

Muller

(eds) Institutional Research in Support of Evidence-Based Decision-Making in Higher Education in Southern Africa. Stellenbosch: SUN Media, pp.331–353.

37.

Prinsloo

(2017) Fleeing from Frankenstein’s monster and meeting Kafka on the way: Algorithmic decision-making in higher education. E-Learning and Digital Media 14(3): 138–163.

38.

Prinsloo P (2019) A social cartography of analytics in education as performative politics. British Journal of Educational Technology 50(6): 2810–2823. DOI: 10.1111/bjet.12872

39.

Prinsloo

Archer

Barnes

, et al. (2015) Big(ger) data as better data in open distance learning. The International Review of Research in Open and Distributed Learning 16: 284–306.

40.

Prinsloo

Slade

(2014) Educational triage in higher online education: Walking a moral tightrope. The International Review of Research in Open and Distributed Learning 14(4): 306–331.

41.

Prinsloo P and Slade S (2017) The elephant in the learning analytics room: The obligation to act. In: LAK’ 17: Proceedings of the Seventh International Learning Analytics & Knowledge Conference, March, Association for Computing Machinery New York, NY, pp.46–55.

42.

Shum

(2017) Black box learning analytics? Beyond algorithmic transparency. In: Keynote presented at the Learning Analytics Summer Institute, Ann Arbor, June 2017.

43.

Siemens G (2010) About. In: The 1st international conference on learning analytics and knowledge, 22 July 2011. Available at: https://tekri.athabascau.ca/analytics/. (accessed 2 September 2019).

44.

Slade

Prinsloo

(2013) Learning analytics: Ethical issues and dilemmas. American Behavioral Scientist 57(1): 1509–1528.

45.

Slade

Prinsloo

Khalil

(2019) Learning analytics at the intersections of student trust, disclosure and benefit. In: Proceedings of the 9th international conference on learning analytics & knowledge, March, pp.235–244. New York: ACM.

46.

Verbert

Drachsler

Manouselis

, et al. (2011) Dataset-driven research for improving recommender systems for learning. In: Proceedings of the 1st international conference on learning analytics and knowledge, February, pp.44–53. New York: ACM.

47.

Williamson

Potter

Eynon

(2019) New research problems and agendas in learning, media and technology: The editors’ wish list. Learning, Media and Technology 44(2): 87–91.

Of ‘black boxes’ and algorithmic decision-making in (higher) education – A commentary

Abstract

Keywords

Introduction

The lure of faster, smarter and more effective

Student data and algorithmic decision-making

Avoiding Frankenstein and Kafka

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References