The race to reimagine education with GenA.I. is on

Nowhere has the disruptive force of generative A.I. been more immediate or profound than in the education sector, which has been rocked by ChatGPT’s ability to generate essays and other homework assignments in seconds using simple, text-based prompts.

Just two months after ChatGPT’s launch, a survey of 1,000 U.S. college students found nearly one-third of them had already used the GenA.I. tool to complete written homework assignments, with almost 60 percent using it to help co-produce more than half of their work.

Teachers, school districts, and higher education professionals around the globe have responded to these new GenA.I. tools with a mix of fear, fascination, and awe. Just days after ChatGPT launched, some prominent school districts and educational institutions rushed to create new policies and ban the use of generative A.I. applications due to their potential to perpetuate fake news, misinformation, plagiarism, and cheating.

Other organizations – including CESAR – are taking a more hopeful view of this watershed moment. In just a few short months, generative A.I. has tipped the balance of knowledge and power toward the people who have access to these tools and know how to use them. Now education leaders are beginning to embrace the opportunity to transform the entire industry for the better.

The birth of a new creative language

J.P. Magalhães, a data science faculty member and principal technical data scientist manager at CESAR, has been studying machine learning for more than 15 years. Up until now, much of  A.I.’s power has been invisible and “under the hood” of the tools and services we use every day. Now, “A.I. code has moved to the world stage and become the main actor,” Magalhães said.

 

Magalhães equates the arrival of generative A.I. to witnessing the birth of a brand new creative language.

“Imagine how much the world changed when we first acquired language. I believe we’re at that same point today with generative A.I. Soon it will be integrated into everything and be as common as the internet or mobile phones. And if you don’t understand how to use this new language, you won’t be able to engage in this world,” – Magalhães said. 

As a scholar and data scientist, Magalhães believes the path forward for generative A.I. shouldn’t be rooted in fear but in intellectual curiosity instead. Much more experimentation with generative A.I. is needed to understand its capabilities and potential impacts, and he cautions against adding barriers where they’re not necessary – particularly within academic settings.

“A.I. is changing every single day. We don’t have books to give to students. We are learning this new language and building this knowledge together with them. When I see a student use generative A.I. in a different way that makes sense and isn’t a copy and paste from its results they are building fluency,” he said.

Helda Barros, coordinator of the professional Master’s Degree in Design at CESAR School agrees. “Educators need to embrace this technology, because it’s already becoming embedded in students’ lives.”

With this in mind, Barros has begun integrating tests and assignments into her curriculum that challenge students to leverage A.I. For example, she recently required students to work with ChatGPT to generate design-lesson topics. In addition to sharing the raw-chat transcript, students had to produce a critical review of the conversation itself, using DALL.E to create a visual representation of their experience.

Magalhães and his undergraduate students are currently studying generative A.I.’s most powerful ability: to extrapolate and fill the gaps in data, leading to new insights not previously possible with classical machine-learning models.

“The results that GenA.I. generates for us are not limited to what we expect. The real power of this technology is that it can generate answers for us that go beyond the questions we originally asked,” he said.

To test this idea, Magalhães and one of his students are researching how A.I. can be used to better understand what doctors are writing in their patient records. “What’s important here is that we are not using A.I. just to identify what the doctors are writing, but also what the healthcare system as a whole is doing with those records. What codes are they using? What are the implications of using them? And how can we optimize the entire process?” he said.

For students, academics, and industry researchers, compiling and analyzing the most up-to-date information on any given subject takes considerable time and creates significant administrative overhead. Generative A.I. tools changed that virtually overnight, according to Barros.

“Any systematic review of the literature as a research method is exhaustive, and data always escapes human eyes. Generative A.I. can be a very helpful tool for professors and students because it effectively gathers information that is relevant to each research question with minimal effort,” – Barros said. 

The devil is in the data

Thanks to generative A.I, anyone with an internet connection and access to ChatGPT now has 24/7 access to a personal tutor that can teach them about anything – paving the way for more student-centered, individualized learning in the future. Teachers and faculty members are now grappling with the need to develop new ways of testing and evaluating their students, and students will have to decide when and how to use A.I. to support their work.

Critics are quick to point out that generative A.I. outputs are only as good as the data that feeds them, and the random data sets used to train today’s most popular tools are inherently biased. Because A.I. development has largely been dominated by groups that reinforce biases already present in society, experts caution that deliberate steps must be taken to decolonize data and facilitate inclusion of marginalized groups as more global data collaborations emerge.

Despite these challenges and high anxiety from A.I. and security experts about privacy, the potential impact of deep fakes, and misinformation, these tools are not going away. Educators are facing a growing obligation to help students and adult learners become “A.I. literate” as more tools and their outputs filter into every corner of daily life. The question is not whether to use these tools for education, but how to do so safely, effectively, and ethically, according to Magalhães.

As generative A.I. evolves and matures, technology developers must take responsibility for what they are creating – and the impact of those creations on society.

“Software developers are not gods. We can’t act like these problems aren’t caused by us. Every line of code I write has my biased point of view in it. When applied to generative A.I., this point becomes extremely important, because that line of code creates a model that can generate billions of responses. As we move forward, I believe we have the capability to use this technology we’re creating to protect ourselves as well – by identifying what is fake news, which images are created by computers, and so on.” – Magalhães said.

Preparing today’s workforce for A.I. innovation

Reskilling today’s workforce to prepare them for A.I.’s impact is on the minds of many business leaders. Within days of the arrival of GPT-4, Goldman Sachs issued a report warning that the latest A.I. systems could automate a quarter of all the work done in the U.S. and Europe, costing some 300 million jobs.

A.I. capabilities are quickly making their way into virtually every business application and process. Organizations that want to thrive in this new world need to look beyond the idea of hiring new prompt engineers and A.I. specialists and consider reskilling their workforces from top to bottom, according to Magalhães.

“It’s not just the leaders and directors who should be learning A.I. Everyone should be. Companies can’t hire specialists and say to them: ‘Make my company data-driven and use A.I. in everything.’ AI specialists are important because they know the process and the models, but it’s the people who make transformation possible, not the technology. It’s the people who will identify the problems and think about possible solutions. And, when they do, they’ll use the right A.I. solutions to solve them,” – said Magalhães.