AI: The Past, The Present and The Unknown Future

Group Submission Type: Formal Panel Session

Proposal

It is commonplace to observe that something called “AI” has infiltrated, to some degree, nearly all domains of cultural production, scientific research, and economic activity in this world. This integration happened so rapidly and thoroughly that the implementation of “AI” has become widespread, particularly in the global context of telecommunications using and reusing its production of content, in all spheres of life and academic activities. If you haven’t used Al knowingly yet, just Google “what is comparative education” and the 1st definition that comes up says on the left ”AI Overview” and in the end “Generative AI is experimental” and invites you to correct or add to the text and this is how the next definition will even better as AI “learns” from your correction. Reading it closely we see that the AI overview does not know the answer to the question of what is comparative education; it only has language about comparative education. This is the major point to keep in mind about generative AI: it now has language it does not (yet) even know. As AI becomes more and more integrated in the daily lives of individuals all over the planet, we must take the initiative to further explore these AI models that are being allowed to control various aspects of our lives.

AI is not software, it is not programmed, it is not Information Technology (IT), it is not Big Data (which combines various data-management technologies), and is not part of other information and communication technologies (ICT), even though it sometimes used in combination of those other technologies. It is a machine learning application, an array of functions, processes, and instruments, produced by a few brand names, right now for free. AI is manufactured by scans of very large amounts of texts as data with the hope of producing knowledge with intelligence, as humans can do easily, but of course it is “artificial” because it is being produced by computer systems. This emergent capability and quality are such that no one anticipated in the past or can predict in the future, as AI is a black box. We know what goes into LLMs (large language models) but we do not know what they exactly are and what will come out.

While in the past human-centered AI teaching and research were accessible only to those fluent in artificial languages, now the major AI breakthroughs come not from manipulating symbols and logic but from surfacing patterns in vast quantities of text. Its programming in natural languages democratizes new ways to answer questions that matter to us, as it exponentially increases the population that AI can reach.
Since the release of ChatGPT, one of the most surprising things about it was how all disciplines, not just language and Generative Education teaching, understand that no question AI research, writing, and translation will ultimately automate some of what faculty typically think of as our domain. In a world of intellectual and creative automation, many white-collar jobs will be at risk of elimination, as many of those are typically a justification for higher education. In some ways, AI has been used to make several processes more time-efficient by removing the need for humans to do calculations or overlook simple processes, but in other ways AI has removed the need in this world for those who previously did these white-collar tasks.

In a comparative education context, this could help identify effective pedagogical approaches for diverse student populations across different cultural and socioeconomic backgrounds. The authors note that the feedback on student assessments could be auto-generated, which could facilitate standardized assessment and comparison of student performance across different education systems. AI-driven intelligent tutoring systems can provide personalized instruction and support students, potentially complementing or supplementing teachers. These systems could be designed and adapted for different educational contexts based on comparative research. The authors note that equity and fairness still remain a challenge. We have to remember that all AI models are biased by their training data. Many leading models show a strong bias for Western values. They can also exhibit new biases created by the attempt to align their outputs using human feedback. Now since it is only in English, English language users can use and “correct” the free AI content generated right now. This question is addressed by Simon Curtis, et al. (2024) who looked at AI in the context of Comparative and International Education, observing that technologies that had exacerbated existing inequalities and perpetuated colonial power dynamics are now used for AI models. Taking those limitations in consideration, overall, AI has the potential to enhance comparative education research and practice, but its responsible and ethical development and deployment require careful consideration, due to its unknown future and unique challenges in rapid adaptation.

The landscape of various professions is expected to undergo radical transformation due to the proliferation of Artificial Intelligence (AI). In this context, Columbia University launched a new initiative to bring together key stakeholders, spanning the private, and public sectors, government, civil society and educational institutions. The initiative’s goal is to envision the evolution of each profession by 2050, considering the current and anticipated capabilities of AI. Several Task Forces have been created to look into specific key topics. Subsequently, the project will develop tailored curricula for universities (starting with Columbia University), ensuring that students benefit so that from preventing the emergence of workers in 2024/2025 who could become obsolete by 2030 because they’ll need significant retraining.

This panel brings together both the theoretical and the practices. From what we can learn from science fiction about the future of AI? What are the key competencies of AI for future education? And, what are some of the “Humanly augmented” AI learning experiences? In discussing the three questions, the panelists will critique current approaches and explore future dimensions. Brings together academia, private sector experts, and students in this panel to explore these dimensions. It will advance Higher Education scholars and researchers thinking and sharing AI and Education's current thematic tensions.

Sub Unit

• General Pool

Organizer

• Radhika Iyengar, Columbia University, climate School

Chair

• Matthew A Witenstein, University of New Mexico

Individual Presentations

• A future of AI: What can we Learn from Science Fiction - Esther E. Gottlieb, The Ohio State University

• “Humanly augmented” AI learning experience - Radhika Iyengar, Columbia University, climate School

Discussant

• Payal Arora, Utrecht University