As educators and administrators consider the implications of ChatGPT and generative AI, one question being asked is “how will educational policies and practices begin to shift?”. Given the fact that generative AI is still in its incubation period, the answer remains to be seen. However, based on early conversations and observations, we’re able to make a few predictions on how we believe is the future of AI in education.
Evolving Academic Integrity Policies and Technology
Technology and academic policy are in lock-step trying to keep up with the evolution of AI. Schools at the forefront of this race are looking towards the future and thinking about how they can reinvest in technology and reimagine their academic integrity policies to keep up with the increasing number of students using generative AI.
Reinvesting in Technology to Prevent vs. Detect
Innovative institutions are viewing “AI detection” as a key component to their response, but not a complete solution. An emerging category of technology providers, like Packback, are developing tools that can spot AI-generated writing, and respond accordingly. Educators are using tools like Packback’s embedded CheckGPT tool, GPTZero, and Turnitin to detect the use of AI generated content in student assignments..
In this new world where students can “copy” from an AI that can generate entirely original text, and even be prompted to “write in a way that sounds human”, the definition of plagiarism changes. Plagiarism is defined as “the practice of taking someone else’s work or ideas and passing them off as one’s own” (University of Oxford) but when copying from AI generated text, a student isn’t copying from someone else — they’re using novel text generated by AI. With ChatGPT calling into question the potential (and the relevancy) to be able to really detect AI-generated content effectively, schools are starting to evaluate if hefty plagiarism detection budgets might be better utilized for more relevant purposes in a post-AI world.
Reimagine Academic Integrity Policies
Schools are also updating their academic honesty policies to account for “plagiarism” from AI-generated text, along with proper citation of generative AI. Unlike text plagiarized from human sources, where evidence of the original source can be found, the only evidence that a piece of text was written by AI are statistical markers about the likelihood of the word patterns. And it is possible for humans to write with similar patterns to AI, leading to false positives. Afterall, these AI models were trained on the text written by humans, making the detection and enforcement of policy even harder. Many schools are understanding that AI generated text will become an acceptable part of students’ writing processes and are evaluating academic integrity policies geared around ensuring students accurately credit their use of AI.
And finally, with AI-generated text being a new phenomenon, many students may not yet understand that copying from these platforms without credit is a form of academic dishonesty. There is an emerging trend towards a more measured hand on potential cases of AI plagiarism, treating initial “violations” as “teaching moments” to train students on proper use & citation of AI generated content, rather than immediately filing an academic dishonesty case for the student.
Evolving Curriculum Design and Learning Objectives
Educators and institutions are shifting toward formative assessments. Rather than the traditional approach of collecting and grading written work, schools are moving toward “authentic” assessments geared toward understanding the process instead of the final product. In just the few months since ChatGPT was released, educators have begun sharing ideas for more authentic approaches, like incorporating more synchronous activities and encouraging students to develop their presentation and collaboration skills. The goal? Create assignments that are engaging, valuable to the learning experience, and actually fun to complete and grade.
How Curriculum Design May Evolve Alongside AI
A few possible ways we believe curriculums may evolve to incorporate more formative assessments include:
- Incorporating self-reflection and metacognition into activities by asking students to reflect on their work, why they made key decisions, and what they would improve.
- Asking students to submit more checkpoints of their work to show their development process from beginning to end, including their brainstorming, research, outlining, editing, and final submission;
- Having computer science students perform code review on ChatGPT-generated code;
- Tasking students with editing and fact-checking an AI-generated essay with known factual and stylistic errors;
- Requiring students to openly use ChatGPT in their written assignments, but documenting where and how they incorporated AI-generated text;
- Asking students to submit the prompt, or set of prompts, they used to achieve a final “output” that meets the assignment criteria, helping them to build prompt engineering skills.
How Learning Objectives May Evolve Alongside AI
Institutions are reevaluating their learning objectives to better prepare students to coexist with AI in the workplace. As AI becomes ubiquitous across education, it’s doing the same in the corporate world — which means that schools need to prepare accordingly. We can anticipate that many schools will rescope their learning objectives to focus on helping students develop the skills needed to succeed in an AI-enabled world. AI text generators like ChatGPT will transform the toolkit and process for many professions, especially ones like copywriting and coding. Models like ChatGPT generate convincing looking text, which can be dangerous since it is not always factually accurate. Additionally, without thoughtful prompting and editing, the output of these models can be extremely generic.
Students will need to develop key skills necessary to successfully collaborate with these tools, including prompt writing, editing and curation skills, fact-checking and correction, and high-level planning skills to develop the objectives for how they want to utilize these tools.
Institutions seeking to be on the forefront of this technological shift may consider adding courses in each discipline that specifically address the use of AI in that field, for example; Generative AI in Journalism, Generative AI in Marketing, or Generative AI in the Creative Arts, etc. Additionally, computer science programs that lack structured coursework around AI are at risk of becoming out of date, with these tools already offering support with automating test writing, code review, and even code writing.
Strategic Adoption of AI in Education to Support Student Success
While it’s easy to paint AI with a broad brush in the wake of ChatGPT, generative AI models are far from the only application of AI that is relevant to an educational setting. In fact, innovative institutions are using AI to classify, recommend, suggest scores, provide instant feedback to students, personalize assignments, and more.
Platforms like ChatGPT use “generative AI;” that is, artificial intelligence that can generate new material (written, artistic, or otherwise). But generative AI models are not the only application of AI that can be useful in the classroom. An emerging approach, called instructional AI, marries the most effective elements of generative AI with pedagogical principles proven to support student learning and growth.
Instructional AI uses student-centered, educational applications of AI to enable more students to succeed while improving the instructor’s quality of life by saving time on grading. With learning loss from the COVID-19 pandemic still top of mind for many in education, Instructional AI partners can play a key role in creating personalized, real-time feedback loops for students that help them develop mastery and build confidence.
In fact, the CLEAR research team at the University of North Texas recently published a journal article on the results of a multiyear study evaluating the impact of AI based curriculum technology at their institution. Their findings showed that instructional AI enriches the quality and breadth of feedback students receive, enables instructors to focus on higher-order feedback, and improves student engagement.
What’s Next for AI in Education?
Whether you like it or not (we do!) AI is the future and education will need to transform with it. What we’ve learned from ChatGPT is that fear and backlash are unproductive (and even counterproductive) ways to engage with such tools. The schools and classrooms that have weathered the storm are those that have instead treated generative AI as a catalyst to rethink their assignments and assessments in ways that help students prepare for a future that is increasingly dependent on the relationship between human and machine. The future is here and our reaction to generative AI will be felt for generations to come.
Click here to learn more about how academia is adapting to generative AI.
