Artificial Intelligence

What the Past Can Teach Us About the Future of AI and Education

Current attitudes toward generative AI hearken back to early skepticism about the impact of the internet on education. Both then and now, technology has created challenges but also opportunities that can't be ignored.

• By Dr. David Wiley
• 04/06/23

In 1998, noted technology critic and historian of automation David Noble published his influential article "Digital Diploma Mills: The Automation of Higher Education," in which he warned about the negative impacts the internet would have on education. His main concern was with the potential effects of "automation" on higher education, describing automation in the educational context as "the distribution of digitized course material online." Noble looked into the future and saw a commercial takeover of higher education in which "the new technology of education, like the automation of other industries, robs faculty of their knowledge and skills, their control over their working lives, the product of their labor, and, ultimately, their means of livelihood."

That same year, I launched the OpenContent initiative, proposing that society should combine the communication and collaboration capabilities of the nascent internet with the emerging open source software movement's innovative approach to copyright licensing in order to create and sustain "open content" — digitized educational materials and other creative works that can be freely copied, modified, and used by everyone. I looked into the future and saw a world in which everyone had access to freely available, openly licensed educational resources that supported learning both in and out of school.

As different as these versions of the future were, they were both correct in predicting how the internet would affect education. On the one hand, open content (years later rebranded "open educational resources") has had a massive impact on access to education. My own organization's website alone has recorded more than 1 billion views of our open content by learners around the world over the past decade. And in the U.S., open content is estimated to have saved college students over $1 billion in required textbook costs. But much of what Noble foresaw has also come to pass. Today, more than half of U.S. college and university faculty have contingent (rather than long-term, tenure track) appointments. And even students have lost control of their own labor, as they are frequently required to provide copies of their homework to companies that "protect academic integrity" before faculty will agree to grade their homework.

Two Sides of the Coin: How Educators Are Thinking About the Impact of Large Language Models (LLMs)

The internet has created both incredible opportunities and complex challenges for teachers, students, and the entire enterprise of education. Twenty-five years later, as we discuss the advent of LLMs like ChatGPT, Bard, and Claude, it feels a lot like it did back in 1998.

Some looking into the future see LLMs and related AI as ushering in an era of even greater democratization of access to information. Instead of the current search paradigm, in which you need to scroll through a list of ads intermingled with links to actual search results, and then visit those links one by one hoping to find an answer, you can simply ask your question and get an answer. This is a dramatic improvement in the way we search for and find information. Just a few years in the future you'll look back on a typical Google search and never believe we actually did things in such a clunky way.

But others looking into the future see LLMs causing the collapse of assessment as we know it. LLMs can not only answer multiple choice questions, they can also explain why the correct answer is correct. Gone are the days when a student had to visit multiple cheating websites hoping to find an uploaded copy of their exam. An LLM can just answer the exam questions for them. LLMs are also able to write passable essays on a wide range of topics, and this capability is certain to improve over time, meaning that even written assignments will need to be reframed and rethought. (Of course, math instructors have been dealing with a version of this issue since the invention of calculators, and more recently, tools like WolframAlpha that can solve math problems and show the step-by-step process.) LLMs will cause something like an assessment crisis in education, forcing educators to take a long, hard look at the role of assessment in both supporting and certifying student learning.

Can LLMs Push Us to Develop Better Education Tools and Systems?

As we're looking to the future, there's another place we must not forget to look: the world outside our colleges and universities. LLMs are already radically changing a wide range of fields from marketing to software development. Even my own field of instructional design has already been impacted, and LLMs are changing the way we do that work inside Lumen. Graduates who have been taught how to leverage LLMs (and other AI tools) to radically improve both their creativity and productivity are going to be in high demand very soon, while those whose faculty worked aggressively to make sure they never used these tools may not be. I don't believe it's a stretch to say that institutional reputations with employers and, consequently, their state legislatures, will increasingly correlate with the degree to which their graduates are prepared to work effectively in this new context.

When we look back on this moment 25 years from now, LLMs and other AI will have altered the face of education as profoundly as the internet did in the 25 years before. And both the advocates and the critics of AI will have been proven right in broad strokes. AI will have created even more incredible opportunities for education as well as even more complex challenges. You can't have one without the other.

But more importantly, LLMs and other AI will have changed the world in which our institutions exist as profoundly as the internet changed the world in the 25 years before — possibly even more so. If our institutions hope to be relevant in the future, they will have to do a better job of leveraging the power of this new wave of technologies in the service of education than they did the internet. The pandemic showed quite starkly that we're still not very good at that yet.