AIED 2021 Workshop- Computational Approaches to Creativity in Educational Technologies

The rise of Industry 4.0 – the proliferation of cyber-physical systems, artificial intelligence, big data, and automation – has turned attention, once again, to the interaction between humans and artificial systems (in simplistic terms, robots). The debate on how humans and robots interact largely centres around the interplay between human and artificial cognition. The human-robot cognition interaction fuels practical inquiries into the formation of high-performing human-robot teams, leveraging robots to enhance human cognition, and the capacity for robots to overtake human cognition. At the heart of these conversations, however, lies a critical question – what does the Future of Work look like? As robots take on more and more tasks previously performed by humans, where does that leave the human worker?

A common argument that forms the basis of the World Economic Forum’s characterisation of 21st century skills is that robots, and their underpinning technologies (e.g. artificial intelligence), are ideally suited to routine, algorithmic tasks. Thus, robots can be designed to replace humans as bookkeepers, drivers, administrative assistants, laboratory assistants, telemarketers, paralegals, and even bartenders. This leaves humans with the jobs that robots cannot do. These uniquely human tasks typically revolve around soft skills, for example: leadership and emotional intelligence, persuasion, and negotiation, and, critical to the present discussion, creativity and complex problem solving.

Nearly 60 years ago, the psychologist Jerome Bruner anticipated a future in which “thinking machines” would take on routine problems (i.e. those that are “well-formed” and “amenable to a unique solution”), leaving creativity, and creative problem solving, as the locus of human dignity and worth. Creativity, in other words, is the bastion of human mental sovereignty over robots. This all-or-nothing view is now common, and rather appealing. We have nothing to fear from this perspective, provided we focus our attention on developing our unique capacity for creativity. Indeed, creativity is becoming an important focus of school curricula across the world for precisely this reason.

The reality of the Future of Work, however, may be somewhat more complex and nuanced. While psychologists may argue that creativity is a uniquely human ability, we see more and more examples of artificial systems that are claimed to be creative. The impact of the Future of Work may therefore not be as clear-cut as some would suggest. Will humans continue to dominate creativity, complex problem solving and related soft skills, or will robots usurp human sovereignty? Or, is there a happy medium in which humans and robots interact productively for a net benefit? To answers these questions, it is first necessary to delve into the nature of human creativity and the potential for artificial systems to replicate this capability.

This keynote will explore many of the questions and issues that underpin the interaction between human and artificial systems with regard to creativity. In particular, how do the existing bodies of knowledge about creativity inform the work that is emerging at the intersection of AI and Education. What do we know, and what do we need to know, to facilitate a successful application of AI and Learning Analytics to the development of creativity in education?

Short Bio

David Cropley is the Professor of Engineering Innovation at the University of South Australia. His research interests span creativity in schools and education, assessing organisational innovation capacity, and the nexus of creative problem solving and engineering.

Dr Cropley is author/co-author of eight books including Femina Problematis Solvendis – Problem-Solving Woman: A History of the Creativity of Women (Springer, 2020), Creativity in Engineering: Novel Solutions to Complex Problems (Academic Press, 2015); and The Psychology of Innovation in Organizations (Cambridge University Press, 2015).

Now a recognised expert in creative problem solving and innovation, Dr David Cropley was a scientific consultant and on-screen expert for the Australian ABC TV Documentaries Redesign My Brain (2013), Life at 9 (2014) and Redesign My Brain, Series 2 (2015).

One of many practices to increase creativity in learners is to make space for reflection, where students (but also professionals) write their reflections as they make sense of experiences. As with all forms of expression, the very act of writing can help give form to an inchoate cloud of ideas. Private journals are important, but in formal educational contexts or professional accreditation submissions, such entries can be a resource for a written submission to read by others. This paper argues for a distinctive role that AI can play in this regard, by holding up a metaphorical mirror with carefully designed feedback to make learners more aware of, and reflective about, their reactions to their experiences, especially ones they have found challenging. We describe a web application that uses Natural Language Processing to annotate accounts of personal responses to challenging experiences, highlighting where the author appears to be reflecting shallowly or deeply. This open source tool is already in use by hundreds of students. Our vision is that such tools cultivate qualities that promote creativity in learning, such as greater self-awareness of one’s assumptions and biases, an openness to new perspectives, and a willingness to change fixed behavioural patterns. When there is a requirement to make personal reflection clearly visible to others, understanding the hallmarks of good reflective writing makes for better communicators with clearer narrative.

Promoting creativity is considered an important goal of education, but creativity is notoriously hard to define and measure. In this paper, we make the journey from defining a formal measure of creativity that is efficiently computable to applying the measure in a practical domain. The measure is general and relies on core theoretical concepts in creativity theory, namely fluency, flexibility, and originality, integrating with prior cognitive science literature. We adapt the general measure for Scratch projects. We designed a machine learning model for predicting the creativity of Scratch projects, trained and evaluated on ratings collected from expert human raters. Our results show that the automatic creativity ratings achieved by the model aligned with the rankings of the projects of the expert raters more than the experts agreed with each other. This is the first step in providing computational models for describing creativity that can be applied to educational technologies and to scale up the benefit of creativity education in schools.

Online collaborations for knowledge creation have assumed a growing role in educational settings. One of the major challenges faced by educators who incorporate such practices is assessing the quality of the collaborative process and its product. As traditional quantitative assessment methods focus on individual performance, this requires the development of new paradigms and methods alike [1]–[3]. This is especially true for higher-level processes, such as creativity. Creativity has long been considered a peak of human capabilities. It derives from the ability to form new meaningful combinations out of available resources. While individual creativity refers to a new mental combination that is expressed in the world, group creativity refers to a product that is created through interaction by a group, a work team, or an ensemble [4]. In online environments, collective creativity can result from multiple interactions between group members and the shared content, that lead to the emergence of novel shared meanings [5], [6]. Interactivity is thus a fundamental requirement for collective creativity, but its principal mark, that differentiates it from a mere aggregation of knowledge, is emergence [7].

Recognizing emergence calls for methods that can adequately describe complex dynamic organizational forms [8]. In line with this and inspired by both theory and findings in the Cognitive Sciences, our research applies network analysis tools such as community detection to identify emergent combinations of content within large online collaborative discussions. The discussions were part of blended graduate-level courses and conducted within a platform that enables participants to both post and link posts. Main findings include the revealing of a latent networked modular structure, and the convergence of content modules, enabled by user-generated links. These represent the formation of novel shared associations, which can be viewed as the basis for collective creativity.

[1] D. J. Leu, C. K. Kinzer, J. Coiro, J. Castek, and L. A. Henry, “New Literacies : A Dual-Level Theory of the Changing Nature of Literacy , Instruction , and Assessment Literacy as Deixis,” J. Educ., no. April, pp. 1–18, 2017, doi: 10.1177/002205741719700202.
[2] B. Csapó, J. Ainley, R. E. Bennett, T. Latour, and N. Law, “Technological Issues for Computer-Based Assessment,” in Assessment and teaching of 21st century skills, vol. 9789400723, 2012, pp. 1–345.
[3] P. Williams, “Assessing collaborative learning: big data, analytics and university futures,” Assess. Eval. High. Educ., vol. 42, no. 6, pp. 978–989, 2017, doi: 10.1080/02602938.2016.1216084.
[4] R. K. Sawyer, Explaining creativity: The science of human innovation. Oxford University Press, 2012.
[5] N. Sher, C. Kent, and S. Rafaeli, “Creativity Is Connecting Things: The Role of Network Topology in Fostering Collective Creativity in Multi-Participant Asynchronous Online Discussions,” in 53rd Hawaii International Conference on System Sciences, 2020, pp. 310–319.
[6] S. Rafaeli and F. Sudweeks, “Networked Interactivity,” J. Comput. Commun., vol. 2, no. 4, p. JCMC243, 1997, doi: 10.1111/j.1083-6101.1997.tb00201.x.
[7] L. Yu, J. V. Nickerson, and Y. Sakamoto, “Collective Creativity: Where we are and where we might go,” in Proceedings of collective intelligence., 2012.
[8] S. Faraj, S. L. Jarvenpaa, and A. Majchrzak, “Knowledge collaboration in online communities,” Organ. Sci., vol. 22, no. 5, pp. 1224–1239, 2011, doi: 10.1287/orsc.1100.0614.

Creative writing has become a playground for AI systems "coup-de-force" battle, with AI developers trying to get their algorithms to produce poems, stories and screenplays at human level (fortunately, it is not really successful so far). Are we heading for a human vs. machine creativity competition? As a screenwriter and an AIED person I want to look at what AI can do and what it can't do in terms of creative writing, and see if and how can a "machine-in-the-loop" approach lead to a breakthrough in creative processes beyond what people or machines "on their own" are capable of. This talk will not be on the computational aspect of creativity, but rather on the point of view of the creative person interacting with an AI system, and it is promised that many of the points will be subjective and debatable.

Creativity and Computational Thinking (CT) have been both extensively researched in recent years. However, the associations between them are still not fully understood despite their recognition as essential competencies for the digital age. We have studied these associations over a few randomized controlled studies, with middle school children in Spain and in Israel. We collected data using standardized creativity test (Torrance's TTCT) and cross-referenced it with log files that documented the students' activities in the Kodetu game-based learning environment for CT. Our research findings indicate some interesting associations between CT and Creativity and the ways in which they could promote each other.

This talk will describe a Computational Creativity Framework for Learning Technologies (CCF-LT). The CCF-LT provides a theoretical model of creativity pedagogies, a computational representation of students’ creative processes for assessment and feedback, and a software architecture for implementing the Framework. A software prototype of the CCF-LT, The Muse, was built to evaluate the Framework and show the practical value of the grounded theoretical Framework. The Framework was evaluated using a cross-over experimental design involving a workshop that taught Scratch programming to tertiary level students who used The Muse to complete two projects.

Our findings show that the CCF-LT is valuable for the teaching, modelling, and assessing of creativity at the tertiary level. The CCF-LT provides a representation of creativity pedagogies using the IMS Learning Design specification, two mappings that support interchangeability of activities based on learning objectives and assessment criteria, and a crucial representation of students’ creative processes, called Student’s Personal Creative Process (SPCPs). The two mappings are based on the Innovation Phase Model (IPM) and the Creative Solutions Diagnosis Scale (CSDS) criteria developed by Cropley & Cropley. Additionally, several creativity factors and conditional logic rules for providing real-time feedback to students on their creative processes are integral to the CCF-LT. The Muse is a system that implements the CCF-LT and supports educational technologists in the practical use of the Framework.

The CCF-LT enables future work in creativity by the provision of a computational platform for exploration of students’ creative processes. This can be achieved by applying Artificial Intelligence techniques such as Machine Learning to the analysis of SPCPs. For example, the resulting repository of SPCPs is a valuable resource for analysis of students’ creativity, such as, determining how to derive successful patterns of students’ creativity, which can then be used to develop new creativity pedagogies that are based on successful prior experiences. In this scenario, further analysis can extract the specific contexts in which such patterns were successful.