With AI-powered tools like ChatGPT making waves across various industries, it's no surprise that its use in the education sector has sparked both excitement and concern.
Misinformation, biased responses, privacy, and the ethics of potential job displacement are common concerns surrounding AI-powered tools. In education specifically, teachers worry about the possibility of students using them to cheat on assignments or write their essays for them.
To combat that, several AI detectors like GPTZero, which claim to be able to detect if something has been written by a human or technology, have hit the market. But even the accuracy of those technologies has been challenged.
Some school districts have taken a hard stance and banned the use of artificial intelligence from their district while others believe in the advantages it poses to both teachers and students, and believe that teaching students how to use the new technology ethically and responsibly is a better approach.
Regardless of which side of the debate you stand on, there’s no doubt that artificial intelligence stands to revolutionize the way teachers work, and the way students learn. And since the accuracy of AI outputs depends largely on the quality of its inputs, learning how to prompt it better will only help improve its efficacy.
The benefits of AI in education are that it eases the workload of teachers by helping them plan lessons, design curriculums, draft administrative communication, and grade coursework. AI can dramatically minimize the time teachers spend preparing to do their best in the classroom.
So let’s take a look at some of the ways AI is currently being used in education and take a look at which strategies are successful and which aren’t.
When Stephen Lockyer, a primary school teacher in west London, shared on his Twitter account how ChatGPT helped him plan three lessons on how volcanoes are formed, the comments section was flooded with conflicting opinions. His prompt was simple;
Plan three lessons to explain how volcanoes are formed. Each lesson needs an introductory activity, information input, a student task and a plenary.
GPT’s output was quite detailed and accurate. It listed everything that he asked for in three different lessons that he could then use in his classroom. And while some argued that they were basic and boring lesson plans, Lockyer argued that it was only intended to be a starting point that he could later add to.
Lockyer’s second prompt to ChatGPT was;
[C]an we tabulate this?
The response – the three lessons in a well-organized table format. Something that could be submitted as a written lesson plan to administrators which is often required of teachers.
So was this an effective use of AI? Ask any teacher what they like and dislike about their job and I can almost guarantee that at the top of their cons list would be the time spent outside of working hours preparing lessons and grading papers.
So if using it to plan lessons can shave hours off their workday and reduce teacher burnout, a growing problem in the US, I’d say so. And as Lockyer later said in an interview, “Your lesson plans are your recipe—you still need a chef,” meaning, you still need a teacher to spruce it up a bit and then deliver it to students.
Chris Mah, a former high school English teacher and PhD student who studies teacher education at Stanford University, also thinks that ChatGPT can save teachers hours of planning, yet his approach is even more refined than Lockyer’s, which may yield an even better result.
He points out that contrasting cases, also known as concept attainment, is a teaching technique used a lot in education. The idea is to offer students both positive and negative examples of a given topic so that they ultimately learn more about the topic.
In fact, natural language processors used in AI bots like ChatGPT are also trained using the concept of contrasting learning. The idea is that by exploring what something is not, we can develop a more comprehensive understanding of what it is.
So Chris used ChatGPT to create specific examples to be used for a contrasting cases lesson.
What’s really great about Chris’ method is the process he has come up with when using ChatGPT;
Process: “Establish criteria, Prompt, Review and adapt, Prompt for contrasting cases, review and Adapt.”
“Reviewing and adapting” GPT’s output is a great way to minimize the possibility of spreading misinformation which is one of the main concerns of educators today. Teachers can make sure that the information is accurate, relevant, and aligned with their specific learning objectives before presenting it to their students.
Here is Chris’ first prompt;
Give me five real-life examples of kinetic energy.
He reviewed the list, found that it gave him 5 good examples of kinetic energy (but would have adapted it if he found that one did not meet his criteria), and then prompted it for contrasting cases;
Now, give me five real-life examples of potential energy.
Once again, he reviewed it and this time decided to adapt it – but he let ChatGPT do that too. He wanted an example that might challenge his students a bit more so he asked;
Now give me an example that involves both types of energy.
Now that he has a solid list of kinetic energy and his contrasting cases, he can use them to teach an engaging physics lesson. He can give them to his students one by one so they can debate them, categorize them, and explain them in more detail.
Chris’ idea of using ChatGPT as an “example machine” is a great way for educators to use AI because it offers targeted and specific prompts that will get high-quality outputs.
Teachers aren’t just stopping at lesson planning. Some have used it to write emails to parents, write letters of recommendation, create grading rubrics, and even grade papers.
Teacher Darin Nakakihara is excited about the amount of time he can save by using ChatGPT to help grade the stack of papers that he constantly has piled on his desk.
Nakakihara took a 5th graders’ essay and prompted ChatGPT;
Read this 5 paragraph essay and provide meaningful feedback for the author.
ChatGPT produced a list of detailed feedback on the structure of the essay which Nakakihara seemed impressed with. Others who have tested its use for grading have said that ChatGPT’s grading capabilities seem inaccurate and are concerned that it can’t account for much creativity.
So if it’s going to be used to offer feedback to students on essays, maybe more specific prompts would help the outcome. Something like, “Does this 5 paragraph essay give an example of why freedom is important from the book ‘Among the Hidden?’”
But Nakakihara’s next prompt was something GPT has a good chance of excelling at;
Identify the grammar and punctuation errors in this five paragraph essay.
AI grammar checkers are nothing new, but they continue to be powerful tools for teachers. By saving himself a significant amount of time, Nakakihara has not only made his own life easier but has also created an opportunity to engage with his students in more meaningful ways, like sitting down with them for some one-on-one feedback.
Let’s take a look at the other side of the coin. Nathan Manousos used ChatGPT to help him practice a foreign language demonstrating the potential for this tool to be used as a learning aid.
Now, ChatGPT does not yet have a talk-to-text feature, so he cleverly used the built-in transcription feature on his Mac to speak to ChatGPT in Mandarin. This is what he said (translated to English);
OK, I want to practice this situation, you can be the shop owner and I'll be the customer, and I've come to your restaurant and we practice this situation, ok?
He then wrote a Greasemonkey script to send ChatGPT’s response directly to Azure, an AI voice generator that turns text into speech, resulting in a two-way conversation in Mandarin.
It would be interesting to learn more details about how he wrote the Greasemonkey script which he promised to elaborate on, but as of now has not. However, even if you don’t turn the responses into voiced responses, writing back and forth still gets you conversation practice in a foreign language.
The problem that arises with this case is, wouldn’t it be just about as accurate as having a conversation with google translate – at best, basic and choppy? Also, since it is a non-native-speaking student using it for practice, checking the accuracy would be pretty difficult to do.
A native mandarin speaker in the comments of Manousos’ Twitter post seemed to confirm these suspicions noting that, “for Chinese native speakers, some sentences sound a little weird.”
So conversing with an AI would not measure up to conversing with a native speaker which most foreign language instructors would advocate as the best way to practice a foreign language.
But it’s still an interesting use of ChatGPT’s conversational capabilities which are beyond those of other AI-driven translators like Google translate.
Whether the impact of AI on education will be positive or negative is still a polarizing debate. As with any new technology, there are bound to be challenges and learning curves, but history has shown us that the best way to approach these challenges is by adapting to them.
When the calculator came along, many of the same concerns were raised by math teachers. And Google was once viewed with suspicion and concern, but today is widely used as an essential tool in education.
The same can be said for artificial intelligence. And while educators must remain vigilant about potential misuse and inaccuracies, they can also explore ways in which it can enhance student learning experiences and their own workflows.
Ultimately, it will be up to our educators to adapt to the ever-changing technological world we live in and to embrace the opportunities that AI presents for the future of education, or not.
AI for Teachers: Curriculum Planning, Lesson Design & Grading
AI in the Classroom: Ways Teachers Use AI tools today