The introduction of artificial intelligence (AI) in education has sparked contrasting approaches among institutions. Following the launch of ChatGPT in late 2022, New York City’s education department swiftly banned the chatbot, citing concerns over its potential to negatively impact student learning and issues regarding content security and accuracy. In stark contrast, Franklin School, a private institution in Jersey City, New Jersey, has embraced AI, integrating it into their curriculum to support teachers and enhance student engagement.
Will Campbell, head of Franklin School, emphasized that the aim of using AI is not to replace educators but to enrich the learning experience for students while providing teachers with more time to focus on instruction. The school, which opened in 2022, has implemented custom chatbots trained on approved course material to serve as tutor-like aids. These AI tools have allowed faculty to delegate routine administrative tasks, thus enhancing their capacity to support students directly.
Evolving AI Education Strategies
At the university level, similar philosophies are emerging. Ethan Mollick, Ph.D., a professor at the Wharton School of the University of Pennsylvania, introduced explicit AI usage guidelines in his syllabus as early as January 2023. He has become a prominent advocate for the integration of AI in education, collaborating with AI companies, including OpenAI, to craft educational guides. His bestselling book, *Co-Intelligence*, explores the transformative potential of AI in learning and work.
Mollick highlights the significant advantages AI can offer, stating, “We have some early evidence that it’s an incredibly powerful teaching tool.” The ongoing debate in educational circles revolves around whether AI can truly replicate the value of human instruction. While some argue that AI cannot replace human teachers, others contend that AI-powered tutoring provides a crucial resource for students lacking access to traditional tutoring.
The Tutoring Gap
Research consistently shows that individualized tutoring significantly enhances learning outcomes. A landmark study by educational psychologist Benjamin S. Bloom in 1984 revealed that students who received one-on-one tutoring could achieve scores up to two standard deviations higher than those taught through conventional methods. This “2 Sigma Problem” underscores the effectiveness of personalized instruction. Yet, logistical and financial barriers prevent many families from accessing such support.
A peer-reviewed study published in March 2024 by the University of Wisconsin-Madison confirmed a correlation between socioeconomic status and tutoring access. The findings indicated that only about 15% of students receive any tutoring, with fewer than 2% obtaining high-quality assistance. Among students who struggle academically, particularly those with grades of C or lower, less than 4% have access to effective tutoring resources.
AI’s Role in Closing Educational Disparities
In theory, AI could bridge this tutoring gap by providing students with accessible, tutor-like resources at home. Advances in natural language processing enable AI tools to engage in conversational interactions, explain complex concepts, and adapt responses similarly to human tutors. As these technologies evolve, they can offer additional support mechanisms, including the ability to analyze photos of student work or conduct live video sessions to address real-time questions.
Many educators, including Michael Hilton, Ph.D., a professor at Carnegie Mellon University’s School of Computer Science, have observed a decline in office hour attendance as students increasingly turn to AI for assistance. “It’s much more of this ‘teach them how to fish’ paradigm in office hours,” Hilton noted, emphasizing that AI can handle basic queries, allowing educators to focus on more complex learning needs.
Despite AI’s potential, concerns remain regarding its efficacy as a replacement for human tutors. A recent report from the Brookings Institution highlighted that the biggest barrier to scaling effective tutoring programs is cost, estimating expenses of $1,000 to $3,000 per student annually for high-impact models. The financial investment required for private tutoring often exacerbates existing disparities in educational achievement.
Human Connection in Tutoring
While organizations like Upchieve, founded by Aly Murray, leverage AI to provide free online tutoring, they underscore that human tutors remain indispensable. Upchieve’s platform integrates AI for certain functions, such as session summaries, but retains a strong focus on human interaction. Murray emphasized, “AI does have lots of exciting applications to education. It’s going to help us accomplish our mission. However, humans are really great for tutoring specifically.”
Research conducted by Upchieve in partnership with Microsoft and the Gates Foundation tested an AI tutoring chatbot, known as UPbot, which was trained on over 70,000 human tutoring transcripts. Despite its capabilities, engagement levels were low, with only 20% of students utilizing the AI tool. In contrast, 92% of tutoring sessions involved human tutors, indicating a clear preference for interpersonal connections.
The Limitations and Risks of AI Tutoring
The effectiveness of AI tutoring remains under scrutiny, as demonstrated by Upchieve’s findings, which showed no significant difference in student learning or confidence between human-only and AI-only sessions. While AI may offer valuable assistance, there is a risk that it could widen existing inequalities. Murray cautioned that affluent students may benefit disproportionately from AI tools, potentially advancing their learning faster than their lower-income peers.
The rapid evolution of technology has historically accompanied disparities in access. For instance, a study using data from the City Health Dashboard and the American Community Survey revealed that high-income neighborhoods enjoy significantly higher rates of broadband access compared to low-income areas. In this context, experts warn that those who can effectively leverage AI will excel, while those who struggle to navigate these tools may fall further behind.
Incremental Progress Towards Equity
Despite these challenges, many scholars agree that while personalized human tutoring is optimal, the more pressing question is whether AI tutoring can provide any support at all. As Campbell pointed out, “If I could put a chatbot that looks like my teachers into a community that doesn’t have the resources…I fundamentally believe that is what we’re meant to be doing for education.”
Educational technology companies are navigating these complex issues as they develop online learning tools. McGraw Hill, founded in 1888, has transitioned from traditional publishing to offering educational content, software, and services, integrating AI cautiously to enhance learning experiences. Chief Data Science & AI Officer Dylan Arena emphasized the importance of maintaining human involvement in the educational process, ensuring that AI complements rather than replaces human teaching quality.
Similarly, Khan Academy has introduced AI components through tools like Khanmigo, designed to support students without simply providing answers. Chief Learning Officer Kristen DiCerbo noted that these tools have been particularly beneficial for English language learners, helping to address access disparities and promote effective learning.
As AI continues to evolve in educational settings, the focus remains on ensuring equitable access to resources that can benefit all students, regardless of their backgrounds. With careful implementation and a commitment to maintaining human connections, AI has the potential to play a transformative role in bridging educational inequalities.
