Michigan State University scholars contributed to research examining current and future use cases of artificial intelligence (AI) across health, education and the workplace. The research also identified future opportunities and challenges in all three sub-areas.
Thirty-three scholars from across the world collaborated on the article. Four are College of Education Spartans: Professor Christine Greenhow, Educational Psychology and Educational Technology (EPET), Ph.D. students Aisel Akhmedova and Jennifer Sutcliffe, and 2024 Curriculum, Instruction and Teacher Education (CITE) alum Selin Akgün, now assistant professor at the University of Minnesota.
“As the world wrestles with the possibilities and pitfalls of applying AI, we were thrilled to collaborate with an international group of scholars – including two of this year’s Nobel Prize winners – to co-author this article, which maps the potential impact of generative AI on education, socioeconomic inequalities and policy-making,” said Greenhow. “We believe this work will be broadly accessible to readers who are excited by or concerned about applications of AI in our lives.”
Opportunities with AI
The MSU scholars contributed most meaningfully to the education section of the paper. The research was published in the June 2024 issue of PNAS Nexus.
Scholars identified “promising” opportunities for AI across the education spectrum, including the increased ability to translate text and homework into different languages and personalize learning experiences. In some previous research, case studies have shown that students received higher test scores when using AI that focused on personalized learning.
AI can also alleviate, expedite or inspire work for teachers. For example, some teachers have utilized platforms to grade assignments.
These are but featured highlights of the opportunities, and others not directly mentioned, in the paper.
“Aspects of generative AI that augment existing teaching practices, such as personalized learning through chatbot tutors, have already demonstrated the capacity to improve learning outcomes in some cases,” said Sutcliffe.
Concerns for AI use in education
However, amidst these optimistic viewpoints, concerns linger.
The paper also highlighted challenges, such as equitable access, understanding of AI applications and possible discriminatory, biased or incorrect outputs created, which could skew experiences and learning.
“While acknowledging the affordances of AI, it is also critical to discuss the source of these ethical concerns,” said Akgün. “The problem stems from how the use and consequences of AI technologies are considered as objective and neutral. However, we know that humans carry historical and systemic social, cultural and political biases; therefore, machines learn from our human biases and inherit them from humans.”
In one sobering example, previous research found humans, generally, are unable to determine whether a text was written by a human or was AI-generated. Moreover, AI-generated text was described by one study as “generat[ing] more convincing misinformation than humans.”
The study also denoted a human-centric challenge. Females report using ChatGPT, a generative AI tool, “less frequently” than male counterparts.
For those who do use AI, regardless of gender, there is also a concern about the reliance on AI platforms. In other words, how much is too much AI usage? At what point does AI take away independent learning opportunities?
Contrarily, does learning how to use AI bolster learning opportunities and competencies?
Thinking ahead
Certainly, questions remain for AI and education, and this study doesn’t aim to resolve them.
However, the research does suggest what comes next in terms of policy and practice.
“[Education stakeholders should] redefine the skills and competencies necessary to effectively utilize generative AI,” wrote the paper’s authors.
To put it in another context, the scholars offer this analogy: “Calculators did not remove the need for students to learn the properties of algebra and develop mathematical reasoning.”
Therefore, the paper continues: “[C]urricula must teach how to successfully describe and share ideas, both with and without assistance from generative AI.”
An argument could be made that AI can reduce independent learning – but the scholars also argue that its integration (particularly for predictive text capabilities) could be used to teach students critical thinking skills. After an AI platform provides an output to a prompt, the students could evaluate the text and keep iterating the prompts to produce clearer, more correct results.
Thus, AI also provides an opportunity to bolster fact-checking competencies.
According to the paper, “among more than 3,000 U.S. high school students and undergraduates, 96% did not know how to evaluate the trustworthiness of websites.”
The scholars suggest improving fact-checking understanding could yield to beneficial results for teachers and students alike.
“Applications of AI are perhaps most successful with a knowledgeable human collaborator who iteratively vets and shapes the outcome,” said Greenhow. “AI has the potential to assist education but only if we understand and address its challenges.”