Welcome to part 2 of Crowdmark’s ‘Rise of ChatGPT’ series, which explores the impact of chatbots on education and grading in 2024
Last time, we covered ChatGPT’s 2023 launch and the rise of AI-driven chatbot technology, along with the general impact of this technology on the higher education space. This time, we’re diving into grading and feedback.
Where can tools like ChatGPT benefit student grading?
In our first post, we noted chatbots have a clear role in the rudimentary work of correcting student text: they catch spelling or grammar mistakes or assess text for gaps. For students calibrating to the expectations of a post-secondary arena, chatbots may be a useful source of initial feedback.
But the amount of text a chatbot can ingest and crunch in a single query is also limited. Depending on who you ask and what tool you’re using, that figure ranges anywhere from 8,000 to 25,000 words.
So, while no one’s easily using ChatGPT to write a coherent thesis-length piece of work, it’s inevitable that chatbot applications will become more robust with time.
One area that’s already getting that lift is computer science.
The code doctor will see your homework now
As with many STEM fields, gaining mastery in software development is a technical and challenging process.
Tools to help coding students produce better work are nothing new. Journalist and developer David Gewirtz has written about existing code debugging tools and ChatGPT’s current abilities and shortcomings for reviewing code, including rewriting regular expression code and finding errors.
TL;DR? It’s early days and chatbots aren’t great at functional code—at least, not yet.
So, submitting raw chatbot code in lieu of doing the assignment isn’t in the best interest of students or the people who grade their work.
However, Amy Haddad, who writes books and tools to help people code in Python, likens reading code for software developers to athletes who watch their game films to improve technique. In her view, doing so teaches you to “see patterns,” “learn new strategies to solve the same problem,” “improve code design,” and get better at “spotting bugs.” And chatbot code may eventually become abundant in that literacy effort.
But the benefits aren’t limited to new developers.
In the corporate world beyond academia, “Some software teams struggle to properly document their code,” says Jamie Gilgen, Crowdmark software engineer. “Documentation helps people new to your team learn how you work, makes code handovers simpler, and ensures consistency and cohesion in what you build. Ultimately, that practice results in better products. But it also requires discipline to do well. Assigning code documentation to a chatbot could be a perfect solution for teams with this pervasive challenge, assuming it’s capable of understanding your needs.”
For now, chatbots may offer a useful first screen for errors before handing in an assignment or in a self-marking exercise.
We’re not going back to the pre-automation world
While ChatGPT’s speed at changing how we work with technology can feel mind-boggling, it’s equally clear that AI-driven automation tools aren’t going away.
“Ignoring chatbot technology won’t make it go away,” says Michelle Caers, CEO of Crowdmark. “Our goal should always be to augment what human beings can do, and not replace the creative work of people. As a software company in the education space, we work to leverage automation that helps people teach and learn in tandem with automation, not in opposition to it.”
Founded in 2012, Crowdmark was created to help educators scale their time by automating the administrative parts of assessment. Using the system, instructors collaborate with their marking teams to bank questions and comments on student work. Over time, they collaboratively build libraries for assignments and exams within each course.
Once assembled, this content can then be reapplied for multiple student cohorts and across semesters, automating the hassle of manually recreating the same assessments. Grading with Crowdmark allows effective comments from individual markers to be shared across the team, lets teams grade handwritten or digital work, and gives instructors more opportunities to assess the grading feedback for overall consistency, ensuring an equitable experience.
Some instructors are going back to paper-based tests or administering assessments via devices without full Internet capacity, the results of which can then be assessed through Crowdmark. Like chatbots, the Crowdmark system can also interpret handwriting and apply its content for other uses.
“In general, educators choose to use Crowdmark to give the richest possible feedback to students. They understand the positive impact that their time and energy can have on someone’s learning journey,” says Caers. “Yet, scribbling out the answer to the same problem thirty times makes few graders happy. Marking can be an endurance exercise, especially for large classes. We want to help graders bring the same rigour and energy to the last paper they mark as they do to the first.”
Protecting copyright and intellectual property remains crucial
It’s also worth noting that in August 2023, U.S. courts signaled that we cannot assume work created with automation software will receive the same protections as what humans generate.
According to Reuters, the judge ruled that, “a work of art generated without human involvement cannot be copyrighted.” The ruling followed a similar case where, also according to Reuters, the U.S. Copyright Office, “rejected an artist’s bid for copyrights on images generated through the AI system Midjourney despite the artist’s argument that the system was part of their creative process.”
Part of the problem lies in the intellectual property folded into language learning modules, much of which may have been gathered and fed to algorithms without the explicit permission of its creators.
Internet artwork is one source of chatbot content ingested without permission. In another notable class action case, authors George R. R. Martin, John Grisham, Jodi Picoult and others are suing Open AI for copyright infringement. The authors argue that the company’s alleged decision to use their work without permission endangers their “ability to make a living.”
It remains to be seen where intellectual property will line up with chatbot outputs.
What do chatbot, copyright and IP trends mean for educators?
As chatbot technology iterates and spreads, it’s clear we’re still in the early cycle of adoption. Fostering experimentation with grading tools is a necessary part of moving with the times, but may carry unforeseen risks.
“It’s been almost twenty years since social networks like Facebook were launched,” says Paul Mitchell, Crowdmark product designer. “While their inventors foresaw some outcomes, many developments—both positive and negative—were entirely unforeseen. Implementing chatbot technology across an organization may also pose similar risks and benefits.”
“Protecting your intellectual property was something we explicitly thought about when building Crowdmark,” says Caers. “Once your course is built, the comments you and your teams spend time creating can be shared and re-used within our closed system in ways that foster collaboration across marking teams and support student learning while protecting your work. Our goal is always to seamlessly reuse and reapply your content while helping you focus your time on activities that foster deeper learning.”
As the technology evolves, properly storing, organizing, and applying comments with improving automation tools will help educators to focus on what matters most—a personalized approach to student learning.
If you are looking to save yourself time grading, while providing richer feedback for students, sign up for a free trial of Crowdmark!