Artificial Intelligence (AI) can benefit many human endeavors and will transform how humans interact and function in many different fields. One such application area where the introduction and entrenchment of AI can be transformative is in education. From primary, secondary, and tertiary schooling to independent adult learners, AI could be interwoven into different ecosystems to help students learn, master, and share knowledge more effectively and efficiently by acting as effective dialog partners and work assistants (Dengel et al., 2021). However, the implementation of AI in education is not conducted in a controlled or systematic manner but is rather occurring asymmetrically across diverse geographies and organizational structures with different policy initiatives (Schiff, 2021). This leaves researchers and system designers with the conundrum of what it means to successfully implement an AI aided ecosystem for learners.

The purpose of this seminar is to explore, discuss, and shape the future of AI within the field of education and learning with regards to its cognitive, physical, perceptual, and societal impacts on humans. Within the education and learning fields, deployment of augmentation tools and systems will likely have a direct impact on the social structures of humanity. We aim to bring experts from Human-Computer Interaction (HCI), Education, Cognitive Psychology, Public Policy, and Philosophy to explore the development of AI educational ecosystems. In particular, we will explore the following interconnected themes:

Achieving Individualized education through AI aided systems: One of the greatest potential uses of an AI education ecosystem would be the fully individualized systems that can tailor content and strategies to an individual’s needs as opposed to a system that optimizes for a majority or plurality of individuals. Learners would then be able to save time and effort when studying by having the optimal routines prepared for them. This could be especially useful for at-risk learners (Yang et al., 2021). The path to achieving a system that works well has some difficulties. Privacy and validation remain a challenge. This is especially true when considering in-the-wild usage. The type of training data that is needed to successfully build a system is difficult to collect and doing so may put participants in studies at disadvantages.

Encouraging knowledge transfer platforms: The transfer of knowledge from individuals to other individuals is difficult to manage on a large-scale platform. Typically, there are two types of platforms: Those which do not consider information redundancy (the same things can be asked and learned repeatedly) and those which seek to create an information corpus (each exchange should be unique). In the former, the information exchange and creation is inefficient and difficult to search. In the latter, the restrictive nature of the exchange makes it difficult to encourage new entrants to the system. In both cases, the exchanges will almost certainly experience power law, where most content is created and maintained by relatively few users, regardless of the assumed incentive system (Vargo, et al. 2016). An AI mediated system brings many opportunities and challenges for creating knowledge sharing platforms. At the same time, the incentive systems needed to make such systems work are unknown.

Trade-offs from AI use: While AI aided systems promise many benefits to its users, it is also necessary to consider if there are negative externalities where the technology may impact the cognitive or perceptual abilities of humans. An example is how pervasive availability of GPS maps on mobile devices and in automobiles have potentially hindered the wayfinding ability of individuals (Ishikawa et al., 2008). While the ability of an individual to quickly find an exact location has been enhanced through technology, the ability to navigate without the device has regressed. In a similar way, technologies that are introduced in education may enhance specific performance but may harm performance outside of the system structure.

Splintering infrastructures in education: Left on its own, and regardless of the intentions of researchers, the asymmetrical implementation of these systems is likely to result in uneven and potentially damaging outcomes for different types of participants (Holmes, et al., 2021). The access to technology and requisite knowledge needed to harvest its benefits are naturally skewed to benefiting wealthy groups, and the systems themselves may have biases baked in which discriminate against different groups. A potential danger is that implementation results in splintering infrastructures, where certain groups continue to use mostly traditional methods of education, while groups with access to technologies experience a beneficial cycle where technologies and systems are continually updated and optimized for their specific needs.

References: Augmented Human and Human-Machine Co-Evolution: Efficiency and Ethics. A. Dengel, L. Devillers, LM Schaal. Frontiers Virtual Reality. 2021.

Ethics of AI in Education: Towards a Community-Wide Framework. W. Holmes, K. Porayska-Pomsta, K. Holstein, E. Sutherland, T, Baker, S, Buckingham Shum, OC. Santos, MT. Rodrigo, M. Cukurova, I. Ibert Bittencourt, KR. Koedinger. International Journal of Artificial Intelligence in Education. 2021.

Wayfinding with a GPS-based Mobile Navigation System: A Comparison with Maps and Direct Experience. T. Ishikawa, H. Fujiwara, O. Imai. A. Okabe. Journal of Enviromental Psychology. 2008.

Education for AI, not AI for Education: The Role of Education and Ethics in National AI Policy Strategies. D. Schiff. International Journal of Artificial Intelligence in Education. 2021.

Editing Unfit Questions in Q&A. A. Vargo, S. Matsubara. International Congress on Advanced Applied Informatics (IIAI-AAI), 2016.

Human-Centered Artificial Intelligence in Education: Seeing the Invisible Through the Visible. SJH. Yang, H. Ogata, T. Matsui, NS Chen. Computers and Education: Artificial Intelligence. 2021.