Generative Ai and Education: A Tool-Based Approach (Part I)

Like in other industries, AI systems will be used in education to automate tasks. In many ways, this process has already started and some aspects of our daily life are meaningfully affected by the use of Large Language Models (LLMs). For example, Grammarly is, in real-time, suggesting alternative means of ordering this sentence. Such a system provides me with timely and actionable feedback to improve the quality of my writing without the need for a human editor to review my work in greater detail (whether a human editor was necessary I leave to you, dear reader). 

A question may arise in considering what I just wrote—will such a system for feedback lead to tangible improvements in my writing or simply serve as a crutch that I rely on? Questions like this will be the source of much debate. I, however, would argue that much of the productive use of Ai, like any other tool, comes out of our intent¹ and an understanding of the unique affordances and boundaries of the technology. 

From building a foundational understanding of the implementation of the technology at first principles, we can better develop policy around the use of Ai in education. Policy and best practices to amplify the positive affordances of our collaboration with the technology while mitigating the drawbacks by steering clear of the boundary conditions beyond which the technology is no longer effective. 

Generative texts and multimedia² will become increasingly common as natural language and diffusion models improve and become more directly accessible as use cases are examined, and the UI to support those use cases are designed. 

Moving beyond considerations and opportunities for automating tasks and exploring avenues for collaboration may yield value. Collaboration between an LLM and a human would necessarily mean a system of checks and balances that puts the human in the loop. What this may look like in the classroom would be a teacher using a modular system like Eduaide.Ai that necessarily puts them in the loop to shape and personalize a generative resource to fit their unique classroom context. This may mean chunking a generated text into smaller subsections with teacher re-writes and headings to meet a student at their present reading level. Or, by taking said generative text and extracting a glossary of keywords to serve as the basis for a vocabulary assignment along with a short direct instruction script co-authored by the teacher and Ai that features two discussion questions based on the reading and presentation that came before. Moreover, such a system may expose teachers to methods and techniques they may not have otherwise considered as the start-up cost for exploring a new and potentially uncomfortable instructional technique is lowered.

In short, the collaboration between an LLM and a teacher would start with stripping away all the technology and asking what it is we hope to achieve. We should not generalize the goals of education here, as they are multivariate and oftentimes context-dependent, but foundationally a goal is in the promotion of learning and the cultivation of the self through that ceaseless journey of learning and self-discovery. We know that learning is an active mental process that involves the acquisition, organization, and use of information in the mind. Since we are humans situated in a society and culture there are a number of social, environmental, cognitive, and technological factors that influence this process. The media we use to support this process also influences it. Therefore, to understand how we use technology to support learning is to understand how we learn. 

This blog entry will be the first in a series to explore the relationship between generative AI and education in a more comprehensive manner. 

Notes

1. What I mean by intent is a combination of knowledge, skills, and disposition with which we approach Ai. That is knowledge of how the system works, skill in wielding the system for our ends, and the dispositions or principles by which we develop our ends in using the system for public good and the alignment of personal and public interests for a greater good (this may be a utopian vision, but one worth pursuing in the development of curriculum around the just use of AI in education). 
2. Generative resources refer to those things that were created in part or in whole with the use of autonomous systems like a large language or diffusion model (ChatGPT, Dalle-2, Midjourney, etc.)