The potential negative effects of AI on education have garnered significant attention. Among college students familiar with ChatGPT, a 2023 survey found that 64 percent had reported using it to complete a written assignment. The increasing use of ChatGPT and similar AI tools has sparked concerns among teachers and parents alike that cheating will proliferate, students will learn less, and interpersonal skills will deplete as technology replaces more in-person interaction. While these concerns are valid, the conversation around AI and education has placed too little attention on the revolutionary role AI is playing for those who aren’t in school at all. By reducing the costs of receiving knowledge, recent AI tools are able to advance the human capital of those who face barriers to educational attainment.
Lowering Transaction Costs
Throughout much of history, receipt of a high quality post-secondary education typically necessitated attending an expensive university, often one in the United States or Europe, leaving those who can ill afford it without access. The last few years stand apart from history as AI substantially reduced the cost to learn. For those in low-income areas, both within the United States and abroad, receiving a high quality education no longer requires moving miles away from home to attend a university, potentially in some remote American town, where the costs of each class are in the thousands of dollars.
Rather, recent advancements in AI, especially with large language models (LLMs), have significantly reduced the barriers to accessing education. Today, AI acts as a tireless personal tutor, patient and adaptable to your specific needs, ready to respond instantly. Whether you seek a course outline, recommended readings, quizzes for knowledge assessment, or constructive feedback from a unique perspective, all you need is an internet connection and a desire to learn.
Thinking on the Margin
Although tools like ChatGPT are not infallible and sometimes provide imperfect answers, assessing the regulatory and policy implications of AI requires thinking on the margin. For those who can’t afford education through traditional means, would they be better off without ChatGPT? For those who struggle to understand their teacher, or for those with less experienced teachers, would they be better off without ChatGPT? For the millions of young girls denied an education because of their gender, would they be better off without ChatGPT?
Free online LLMs have dramatically reduced the cost to learn, helping overcome the persistent disparities that characterize a lack of education access. For example, see Open AI’s exert of how some teachers are using AI to overcome language barriers:
Dr. Anthony Kaziboni, the Head of Research at the University of Johannesburg, teaches students who mostly don’t speak English outside of the classroom. Kaziboni believes that command of English is a tremendous advantage in the academic world, and that misunderstandings of even small details of English grammar can hold back students from recognition and opportunity. He encourages his students to use ChatGPT for translation assistance, to improve their English writing, and to practice conversation.
With increased access to education, individuals are able to raise their productive potential, leading to numerous benefits in the labor market.
Acknowledging AI's Limitations
To illustrate how free online LLMs like ChatGPT can bolster individual knowledge, I solicited its best counterarguments to my perspective. Below are its critiques alongside my responses:
Quality and Depth of Education: One potential criticism might be questioning whether AI can match the depth and quality of traditional education methods, including the nuances of human interaction, debate, and personalized feedback that shapes critical thinking and complex understanding.
Although AI may not rival the quality of a great in-person teacher, again it’s important to think on the margin. Having a teacher of ChatGPT’s quality is better than having no teacher at all either due to financial constraints, labor market discrimination, or other barriers. In essence, receiving some level of education through AI is preferable to receiving none at all.
Digital Divide: While AI lowers barriers to education, the digital divide remains a significant hurdle. Access to reliable internet and technology is not uniformly distributed globally, which might limit the potential benefits of AI education services for the most marginalized.
I agree. The opportunity cost from lacking internet access grows with the value of AI.
Dependency and Critical Thinking: There's a risk that over-reliance on AI for learning could impact the development of independent research and critical thinking skills. Learning to question and research beyond AI-provided answers is crucial.
From my perspective, I believe more concern is due to an over-reliance on in-person learning and feedback. AI's lack of personal bias and its impartiality allows some users to be more comfortable receiving feedback, enhancing learning potential. Consider those many individuals who are not exposed to diverse viewpoints, what their learning potential could be from escaping their echo chamber. With a desire to learn, these gains can be realized through utilizing large language models.
Data Privacy and Bias: The use of AI in education raises concerns about data privacy and the potential for embedded biases in AI algorithms, which could inadvertently perpetuate inequalities or provide skewed educational content.
On the margin, AI appears no more biased than the human teachers we interact with everyday. Consider simple human biases, such as fluctuations in teaching quality due to energy levels. At least these do not afflict an LLM, which can operate without fatigue or anger.
An Evolving Narrative
In conclusion, the narrative surrounding AI in education must evolve to recognize its potential as a tool that, while not without flaws, offers unprecedented opportunities for learning and growth, especially for those disadvantaged groups.
Consider those facing discrimination in educational access and the labor market. In regions where women are barred from formal education and discouraged from pursuing knowledge, AI may prove the most beneficial. By reducing the costs of learning, new technological advancements may offer a pathway to enlightenment and empowerment where traditional routes are closed.