We’ve all heard by now of ChatGPT, the large language model-based chat bot that can seemingly answer most any question you present it. What if there were a way to provide this functionality to students on their learning management system, and it could answer questions they had about course content? Sure, this would not completely replace the instructor, nor would it be intended to. Instead, for quick course content questions, a chatbot with access to all course materials could provide students with speedy feedback and clarifications in far less time than the standard turnaround required through the usual channels. Of course, more involved questions about assignments and course content questions outside of the scope of course materials would be more suited to the instructor, and the exact usage of a tool like this would need to be explained, as with anything.
Such a tool could be a useful addition to an online course because not only could it potentially save a lot of time, but it could also keep students on the learning platform instead of using a 3rd-party solution to answer questions as is the suspected case right now with currently available chatbots.
To find out what this would look like, I researched a bit on potential LLM chatbot candidates, and came up with a plan to integrate one into a Canvas page.
Disclaimer! This is simply a proof of concept, and is not in production due to certain unknowns such as origin of the initial training data, CPU-bound performance, and pedagogical implications. See the Limitations and Considerations section for more details.
How it works
The main powerhouse behind this is an open source, Large Language Model (LLM) called privateGPT. privateGPT is designed to let you “ask questions to your documents” offline, with privacy as the goal. It therefore seemed like the best way to test this concept out. The owner of the privateGPT repository, Iván Martínez, notes that privacy is prioritized over accuracy. To quote the ReadMe file from GitHub:
100% private, no data leaves your execution environment at any point. You can ingest documents and ask questions without an internet connection!
privateGPT, at the time of writing, was licensed under the Apache-2.0 license, but during this test, no modifications were made to the privateGPT code. Initially, when you run privateGPT, train it on your documents, and ask it questions, you are doing all of this locally through a command line interface in a terminal window. This obviously will not do if we want to integrate it into something like Canvas, so additional tools needed to be built to bridge the gap.
I therefore set about making two additional pieces of software: a web-interface chat box that would later be embedded into a Canvas page, and a small application to connect what the student would type in the chat box to privateGPT, then strip irrelevant data from its response (such as redundant words like “answer” or listing the source documents for the answer) and push that back to the chat box.
A diagram showing how the front-end of the system (what the user sees) interacts with the back-end of the system (what the user does not see). Self-creation.
Once created, the web interface portion, running locally, allows us to plug it into a Canvas page, like so:
A screenshot showing regular Canvas text on the left, and the chat box interface on the right, connected to the LLM.
Testing how it works
To begin, I let the LLM ‘ingest’ the Ecampus Essentials document provided to course developers on the Ecampus website. Then I asked some questions to test it out, one of which was: “What are the Ecampus Essentials?”
I am not sure what I expected here, as it is quite an open ended question, only that it would scan its trained model data and the ingested files looking for an answer. After a while (edited for time) the bot responded:
A video showing the result of asking the bot “What are the Ecampus Essentials?”
A successful result! It has indeed pulled text from the Ecampus Essentials document, but interestingly has also paraphrased certain parts of it as well. Perhaps this is down to the amount of text it is capable of generating, along with the model that was initially selected.
A longer text example
So what happens if you give it a longer text, such as an OpenStax textbook? Would it be able to answer questions students might have about course content inside the book?
To find out, I gave the chatbot the OpenStax textbook Calculus 1, which you can download for free at the OpenStax website. No modifications were made to this text.
Then I asked the chatbot some calculus questions to see what it came up with:
Asking two questions about certain topics in the OpenStax Calculus 1 book.
It would appear that if students had any questions about mathematical theory, they could get a nice (and potentially accurate) summary from a chatbot such as this. Though this brings up some pedagogical considerations such as: would this make students less likely to read textbooks? Would this be able to search for answers to quiz questions and/or assignment problems? It is already common to ask ChatGPT to provide summaries and discussion board replies, would this bot function in much the same way?
Asking the chatbot to calculate things, however, is where one would run into the current limitations of the program, as it is not designed for that. Simple sums such as “1 + 1” return the correct answer, as this is part of the training data or otherwise common knowledge. Asking it to do something like calculate the hypotenuse of a triangle using Pythagorus’ theorem will not be successful (even using a textbook example of 32 + 42 = c2). The bot will attempt to give an answer, but its accuracy will vary wildly based on the data given to it. I could not get it to give me the correct response, but that was expected as this was not in the ingested documentation.
Limitations and Considerations
OK, so it’s not all perfect – far from it, in fact! The version of privateGPT I was using, while impressive, had some interesting quirks in certain responses. Responses were never identical either, but perhaps that is to be expected from a generative LLM. Still, this would require further investigation and testing in a production-ready model.
How regular and substantive interaction (RSI) might be affected is an important point to consider, as a more capable chatbot could impact the student-instructor Q&A discussion board side of things without prior planning on intended usage.
A major technical issue was that I was limited to using the central processing unit (CPU) instead of the much faster graphics processing unit (GPU) used in other LLMs and generative AI tools. This meant that the time between the question being sent and the answer being generated was far higher than desired. As of writing, there appears to be a way to switch privateGPT to GPU instead, which would greatly increase performance on systems with a modern GPU. The processing power required for a chatbot that more than one user would interact with simultaneously would be substantial.
Additionally, the incorporation of a chatbot like this has some other pedagogical implications, such as how the program would respond to questions related to assignment answers, which would need to be researched.
We also need to consider the technical skill required to create and upkeep a chatbot. Despite going through all of this, I am no Artificial Intelligence or Machine Learning expert; a dedicated team would be required to maintain the chatbot’s functionality to a high-enough standard.
Conclusion
In the end, the purpose of this little project was to test if this could be a tool students might find useful and could help them with content questions faster than contacting the instructor. From the small number of tests I conducted, it is very promising, and perhaps a properly built version could be used as a private alternative to ChatGPT, which is already being used by students for this very purpose. A major limitation was running the program from a single computer with consumer components made 3 years ago. With modern hardware and software – perhaps a first-party integrated version built directly into a learning management system like Canvas – students could be provided with their own course- or platform-specific chatbot for course documents and texts.
If you can see any additional uses, or potential benefits or downsides to something like this, leave a comment!
ChatGPT has been featured everywhere recently, with some educators being particularly worried because of its ability to craft essays, provide answers to quiz questions, and bypass plagiarism detection software. Teachers are trying to find ways to detect AI generated content, but are we being too hasty to associate AI in education with academic misconduct? Could the tool being used to cheat on courses actually help build them? I put this to the test by asking ChatGPT some questions.
Note: The purpose of this was to see how much of the course development process could be generated by the chatbot.
How to begin?
Let’s start at the beginning: how do we design a course? Well, keeping the process brief for this post, we want to start with some course learning outcomes, align them with some assessments, and generate some activities to help practice for those assessments. I have a background in linguistics and language education, so I wanted to see how this chatbot would respond to some prompts to create a generic English language course.
Chat Transcript
The following is the chat transcript from ChatGPT verbatim. Screenshots of the conversation are available below. Anything under the “ChatGPT” was written by the AI, with my comments in the “Comments” subsection under that.
Question 1
Gathering the learning outcomes
Could you generate 4-5 course learning outcomes for an intermediate English as a second language course?
ChatGPT Answer
Students will be able to understand and use complex grammatical structures in spoken and written English, including verb tenses, modals, and conditional sentences.
Students will be able to participate in discussions and debates on a variety of topics, using appropriate vocabulary and idiomatic expressions.
Students will be able to read and comprehend authentic English texts, such as news articles, short stories, and literary works, and be able to analyze and interpret them.
Students will be able to write clear, well-organized, and grammatically correct essays, using appropriate transitional words and phrases to develop their ideas.
Students will be able to give presentations in English, using appropriate pronunciation, intonation, and stress to convey their message effectively.
Comments
These are standard course learning outcomes for an ESL course. As I simply asked for learning outcomes with no context, the chatbot has no idea how detailed they should be, or how long it will take to achieve these outcomes. Though that was tested in the next question.
Question 2
Basic weekly topics
Using those outcomes, what would the topics for a 10-week course look like?
ChatGPT Answer
Week 1-2: Review of basic grammatical structures, introduction to verb tenses, including present perfect and past continuous.
Week 3-4: Introduction to modals, including can, could, may, might, must, should, and will. Practice using them in context.
Week 5-6: Conditional sentences, including zero, first, second, and third conditionals. Practice using them in context.
Week 7-8: Vocabulary building and idiomatic expressions. Practice using them in discussion and debates.
Week 9-10: Reading comprehension and analysis. Practice with a variety of texts and interpreting them.
Week 11-12: Writing skills, including clear organization, development of ideas and use of transitional words and phrases.
Week 13-14: Presentation skills, including pronunciation, intonation and stress.
Note: These topics might vary depending on the level of proficiency of the students in the class and the specific course design.
Comments
Interestingly here, the chatbot gave me 14 weeks of topics rather than the 10 I asked for. I believe this is because of the structure of my question, because this is fixed in a later question. This was the only glaring mistake during our entire “conversation”.
Question 3
Assessments
What kinds of assessments could be used for these outcomes and topics?
ChatGPT Answer
Grammar and vocabulary quizzes or tests to assess students’ understanding and mastery of grammatical structures, verb tenses, modals, and idiomatic expressions.
Oral presentations, debates, or discussions to assess students’ ability to participate in spoken interactions using appropriate vocabulary and idiomatic expressions.
Written assessments such as essays, short answer questions or summary writing to assess students’ ability to understand, analyze and interpret texts and convey their ideas in writing.
Reading comprehension quizzes or tests to assess students’ ability to understand and interpret authentic English texts.
Writing assessments such as essays, reports, or research papers to evaluate students’ ability to organize their thoughts, develop their ideas and use appropriate transitional words and phrases.
Oral presentations or speeches to assess students’ ability to convey information effectively using appropriate pronunciation, intonation, and stress.
Self-reflection or peer-evaluation could be used as well to evaluate students’ progress and areas of improvement.
Comments
A fairly wide range of skills being tested here – which is also down to my extremely generic requirement of ‘intermediate english language”. I thought the assessments here contained enough that someone designing a course could pick a few of them from each set provided.
Question 4
Alignment
Can you map the above assessments to the previously mentioned course learning outcomes? How would they align?
ChatGPT Answer
Sure, here’s how the assessments could align with the previously mentioned course learning outcomes:
1. Grammar and vocabulary quizzes or tests align with the outcome of being able to understand and use complex grammatical structures in spoken and written English, including verb tenses, modals, and conditional sentences.
2. Oral presentations, debates, or discussions align with the outcome of being able to participate in discussions and debates on a variety of topics, using appropriate vocabulary and idiomatic expressions.
3. Written assessments such as essays, short answer questions or summary writing align with the outcome of being able to read and comprehend authentic English texts, such as news articles, short stories, and literary works, and be able to analyze and interpret them, and also the outcome of being able to write clear, well-organized, and grammatically correct essays, using appropriate transitional words and phrases to develop their ideas.
4. Reading comprehension quizzes or tests align with the outcome of being able to read and comprehend authentic English texts, such as news articles, short stories, and literary works, and be able to analyze and interpret them.
5. Writing assessments such as essays, reports, or research papers align with the outcome of being able to write clear, well-organized, and grammatically correct essays, using appropriate transitional words and phrases to develop their ideas.
6. Oral presentations or speeches align with the outcome of being able to give presentations in English, using appropriate pronunciation, intonation, and stress to convey their message effectively.
7. Self-reflection or peer-evaluation align with the outcome of evaluating students’ progress and areas of improvement.
It’s important to note that the assessments should be designed to be both formative and summative, in order to provide ongoing feedback to the student and help them to track their progress and identify areas for improvement.
Comments
Alignment of assessments to outcomes is one of the most important parts of designing an online course. I was impressed with the combination of outcome to assessment as explaining why an assessment is being given to students is quite common.
Question 5
How could multimedia help students?
Can you think of how multimedia could be used in this course to promote effective language learning?
ChatGPT Answer
There are several ways that multimedia could be used in an intermediate English as a second language course to promote effective language learning:
Video and audio recordings of native English speakers can be used to expose students to authentic English, including different accents and dialects. This will help students to improve their listening comprehension and pronunciation skills.
Interactive online exercises and activities, such as quizzes, games, and simulations can be used to provide students with opportunities to practice and reinforce their grammar, vocabulary, and idiomatic expressions in a fun and engaging way.
Online discussion boards, chat rooms, and virtual meetings can be used to provide students with opportunities to interact with their classmates and teachers in real-time, thus promoting their speaking and writing skills.
Online videos, documentaries, and TED talks can be used to expose students to a variety of interesting and engaging topics, and to provide students with opportunities to practice their listening comprehension and critical thinking skills.
Online flashcards, quizzes, and other tools can be used to help students to memorize and retain new vocabulary words, and to improve their spelling and word recognition skills.
Online or offline multimedia-based learning materials, such as interactive e-books, mobile apps, and podcasts, can be used as additional resources for students to explore and practice their language skills on their own time.
Online resources such as subtitles for films, TV shows or podcasts can be used to improve students’ listening comprehension, vocabulary and grammar skills.
Overall, multimedia can be a great tool to make language learning more interactive, engaging, and personalized, and can be used to support the traditional teaching methods and assessments.
Comments
The multimedia selection here is fairly common in ESL classrooms. Selecting a combination of these would enhance the learning experience. It is also helpful that an explanation of purpose is given for each item on the list.
Question 6
Diversity and inclusion
Students on this course will be from diverse backgrounds and cultures. Can you anticipate any topics that would best be avoided?
ChatGPT Answer
When teaching students from diverse backgrounds and cultures, it’s important to be sensitive to cultural differences and to avoid topics that may be offensive or controversial. Some potential topics that might be best avoided in an intermediate English as a second language course include:
Religion: It’s best to avoid discussing religious beliefs, practices, or controversies, as they can be deeply personal and may offend some students or create unnecessary tension among the class.
Politics: Avoid discussing or debating controversial political issues, as they may be divisive and could create tension among the class.
Cultural stereotypes: Avoid making generalizations or negative comments about any particular culture or group of people, as this can be offensive and perpetuate stereotypes.
Racism and discrimination: Avoid discussing these topics in a way that might make any student feel targeted, uncomfortable or offended.
History: Some historical events can be sensitive and can cause discomfort to some students. It is important to be aware of different perspectives and be sensitive.
It is important to remember that every student is unique and has their own personal experiences and perspectives, and topics that may be acceptable for one student may not be for another. It’s best to use common sense, and if you are unsure, it’s always good to ask your students if they are comfortable discussing a particular topic.
It’s also important to create a safe and inclusive environment, where all students feel respected and valued, and where they can express themselves freely without fear of judgement or discrimination.
Comments
I included this question because of the importance of diversity and inclusion in the classroom, and with an ESL course in particular, students will be from diverse backgrounds. It was good to see that the chatbot could provide not only a list of topics to answer my question but also to give additional advice on the learning environment.
Question 7
Converting this into an online course
How could everything mentioned so far be developed as an online, asynchronous course? Student interaction might be limited to discussion boards as they are likely to be in other countries with different time zones.
ChatGPT Answer
An intermediate English as a second language course can be developed as an online, asynchronous course by using a variety of tools and resources to create a self-paced and interactive learning experience. Some potential strategies include:
Use a Learning Management System (LMS) such as Canvas, Blackboard, or Moodle to create a centralized location for all course materials, assignments, and communication. This will provide students with easy access to all the resources they need to complete the course. Use video and audio recordings, as well as interactive online exercises and activities to provide students with opportunities to practice and reinforce their grammar, vocabulary, and idiomatic expressions in a fun and engaging way.
Use discussion boards, chat rooms, and virtual meetings to provide students with opportunities to interact with their classmates and teachers in real-time, thus promoting their speaking and writing skills. These tools can also be used to create a sense of community among the students, despite their geographic distance and time zones.
Create and provide online multimedia-based learning materials, such as interactive e-books, mobile apps, and podcasts, as additional resources for students to explore and practice their language skills on their own time.
Use online quizzes, tests and assessments to evaluate student’s understanding, and use the results to provide them with feedback and guidance on areas where they need to improve.
Use Self-reflection or peer-evaluation to evaluate students’ progress, and areas of improvement.
Provide detailed and clear instructions for all assignments, activities, and assessments, and make sure that the deadlines are flexible enough to accommodate students in different time zones.
Encourage students to take an active role in their own learning by regularly checking in with them, providing them with feedback, and giving them opportunities to reflect on their progress and set goals for themselves.
Overall, an online, asynchronous course can be designed to be flexible, interactive, and engaging, and can provide students with opportunities to learn and practice their language skills at their own pace and on their own schedule.
Comments
I work in the field of online and hybrid learning, and often adapt existing in-person courses to online versions. I wanted to see if the chatbot could also do this, given that all of the previous answers were based on the assumption of an in-person class. This was not the fault of the ChatGPT – I never specified whether I wanted it to be an online-specific course. An important lesson here is that imprecise input will not result in precise output (see: GIGO for more information, or this article on tailoring your prompts for ChatGPT).
Still, given the updated prompt, it was able to provide both synchronous and asynchronous activities to accommodate for differences in time zones and locations.
Question 8
Asking directly for an opinion
Can you think of anything else that might be important to consider when building this course?
ChatGPT Answer
Here are a few additional things that might be important to consider when building an intermediate English as a second language course:
Assessing students’ prior knowledge and language proficiency: It’s important to assess students’ prior knowledge and language proficiency before the course begins, in order to tailor the course content to their needs and to ensure that they are placed in the appropriate level.
Differentiating instruction: Since students in an intermediate English as a second language course will have varying levels of proficiency, it’s important to differentiate instruction and provide multiple means of representation, expression, and engagement to meet the needs of all students.
Scaffolding and modeling: Provide students with the necessary support, guidance and modeling to complete the activities and assignments.
Provide opportunities for students to practice and apply their language skills in authentic and meaningful contexts.
Encourage students to use English as much as possible: Encourage students to use English as much as possible, both inside and outside of class, by providing them with opportunities to interact with native speakers, read authentic English texts, and listen to English-language music and podcasts.
Use technology to enhance language learning: Use technology to enhance language learning by providing students with access to online resources, such as grammar exercises, vocabulary lists, and pronunciation guides.
Provide regular and constructive feedback: Provide regular and constructive feedback on students’ progress and areas for improvement. This can be done through formative assessments, individual conferences, or through online tools such as rubrics and self-reflection.
Be flexible and responsive: Be flexible and responsive to students’ needs and be open to suggestions and feedback to improve the course.
Cultural sensitivity: Be culturally sensitive, and be aware of the cultural backgrounds of the students and make sure that the course content is inclusive and respectful of different cultures.
Keep in mind that the course is not just about language, but also about culture. Provide students with opportunities to learn about the cultures of English-speaking countries in order to help them to understand and appreciate the language and its context better.
Comments
For this question, I wanted to see if the chatbot could provide additional considerations for teaching and learning. These are all great suggestions, which is probably why the AI selected them – they likely appear in many examples of educational content that it had been trained on.
Question 9
Rewriting course learning outcomes to be more measurable
Can you take the aforementioned course learning outcomes and rewrite them with strong, measurable verbs? For example, “understand” is not easy to measure.
ChatGPT Answer
Sure, here are the course learning outcomes rephrased with strong, measurable verbs:
Students will be able to identify and use complex grammatical structures in spoken and written English, including verb tenses, modals, and conditional sentences with 80% accuracy on grammar quizzes and tests.
Students will be able to initiate and participate in discussions and debates on a variety of topics, using appropriate vocabulary and idiomatic expressions with 70% accuracy on oral presentations, debates, or discussions.
Students will be able to read and comprehend authentic English texts, such as news articles, short stories, and literary works, and analyze and interpret them with 80% accuracy on reading comprehension quizzes or tests.
Students will be able to write clear, well-organized, and grammatically correct essays, using appropriate transitional words and phrases to develop their ideas with 80% accuracy on writing assessments such as essays, reports, or research papers.
Students will be able to give presentations in English, using appropriate pronunciation, intonation, and stress to convey their message effectively with 80% accuracy on oral presentations or speeches.
These objectives are now specific, measurable and have a clear criteria of success. They are also aligned with the assessments and activities that you have previously mentioned.
Comments
I conducted an experiment which mimics the process that instructional designers routinely encounter when designing online courses. There are often a set of learning outcomes given at the start of the development process which regularly use verbs that are heard to measure such as “understand” or “know”. This makes it difficult to align assessments to outcomes because what does it really mean to “understand” something? How can we apply a rubric or grade to it?
Note that, in a real development beginning from scratch, we would want to design the course learning outcomes to be measurable from the start before anything else. In practice, however, I usually work with faculty on courses with predetermined outcomes that are either modified to more measurable ones, or linked to more measurable weekly learning outcomes.
ChatGPT modified the one verb I called out – “understand” – but left “comprehend”. Perhaps this also required a specific mention in the initial question, and is another example of how your initial prompt influences the results.
Question 10
Creating a weekly schedule
Can you make a weekly schedule, from week 1 to week 10, with an overview of activities for that particular week, incorporating the aforementioned assessments where appropriate and aligning all of it with the course learning outcomes?
ChatGPT Answer
Sure, here’s a possible weekly schedule for an intermediate English as a second language course:
Week 1:
Overview of the course and introduction to the course learning outcomes
Assessment of students’ prior knowledge and language proficiency
Introduction to verb tenses, including present perfect and past continuous
Interactive online exercises and activities to practice verb tenses
Vocabulary building and idiomatic expressions
Week 2:
Review and reinforcement of verb tenses
Quiz or test on verb tenses
Introduction to modals, including can, could, may, might, must, should, and will
Practice using modals in context
Week 3:
Review and reinforcement of modals
Quiz or test on modals
Introduction to conditional sentences, including zero, first, second, and third conditionals
Practice using conditional sentences in context
Week 4:
Review and reinforcement of conditional sentences
Quiz or test on conditional sentences
Vocabulary building and idiomatic expressions
Practice using vocabulary and idiomatic expressions in discussions and debates
Week 5:
Reading comprehension and analysis
Practice with a variety of texts and interpreting them
Writing skills, including clear organization, development of ideas and use of transitional words and phrases
Week 6:
Review and reinforcement of writing skills
Writing assessment such as essays, reports, or research papers
Presentation skills, including pronunciation, intonation and stress
Week 7:
Review and reinforcement of presentation skills
Practice oral presentations or speeches
Self-reflection or peer-evaluation
Vocabulary building and idiomatic expressions
Week 8:
Reading comprehension and analysis
Practice with a variety of texts and interpreting them
Writing skills, including clear organization, development of ideas and use of transitional words and phrases
Week 9:
Review and reinforcement of all language skills
Final assessment, which can include a combination of quizzes, tests, essays, oral presentations, or other forms of assessment.
Feedback on final assessment
Week 10:
Review of key concepts and skills
Feedback on the course, and suggestions for improvement
Final self-reflection or peer-evaluation
Note: This is just a sample schedule, and it can be adjusted depending on the specific needs of the students and the course design. Also, it’s important to remember that the assessments should be designed to be both formative and summative, in order to provide ongoing feedback to the student and help them to track their progress and identify areas for improvement.
Comments
The language I used in this question yielded the correct number of weeks from the AI. Each week is structured to include a review of the previous week’s content, which is a common teaching technique. This is asking a lot from the chatbot! It managed to incorporate most of my requests though, which again is impressive.
Question 11
Academic Integrity
What measures could be taken to ensure academic integrity in the online version of this course?
ChatGPT Answer
Academic integrity is a critical aspect of any course, and it’s important to take measures to ensure that students are completing their work independently and honestly. Here are a few measures that could be taken to ensure academic integrity in an online version of an intermediate English as a second language course:
Use plagiarism detection software: Use plagiarism detection software, such as Turnitin, to check students’ work for originality. This will help to detect instances of plagiarism and ensure that students are completing their own work.
Provide clear guidelines on academic integrity: Provide students with clear guidelines on academic integrity, including definitions of plagiarism, cheating, and academic dishonesty.
Monitor online activities: Monitor online activities, such as discussion board posts, chat room conversations, and virtual meetings, to ensure that students are participating in the course honestly and independently.
Use proctoring software: Use proctoring software, such as ProctorU or Respondus Monitor, to monitor students while they take online quizzes, tests, or exams.
Use randomized questions: Use randomized questions, or different versions of the same test to prevent students from sharing answers with each other.
Encourage self-regulation: Encourage self-regulation, and make it clear that the students are responsible for maintaining academic integrity in their work.
Create a culture of integrity: Create a culture of integrity in the class, by emphasizing the importance of honesty, trust, and responsibility.
Have clear consequences: Have clear consequences for academic dishonesty, and communicate them to the students.
Overall, it’s important to establish clear expectations for academic integrity and
ChatGPT froze here, the servers were experiencing high volumes of traffic.
Comments
Again, these are solid solutions which are used in both in-person and online courses. A quick web search would yield similar results as these are shared across most educational institutions.
Thoughts on further usage
It is clear that AI is set to change the way we interact with technology in education (or already has depending on how much you are using it). For this test, I simply provided the most generic input to see what the chatbot would generate. For an intermediate (however that was interpreted by the chatbot) ESL course, the learning outcomes were solid, once measurable outcomes were requested, and I was impressed by the generated weekly schedule including high level overviews of activities and assessments.
With more precise input requirements for course content and outcomes, one would expect an even more accurate output.
There are many ways that this kind of technology can be used to enhance courses – and I can see it being a very useful tool for quick questions such as potential assignments for certain learning outcomes, or figuring out if an assignment prompt might result in unintended consequences. For example, see the following:
Bonus Question
What could go wrong?
Can you think of any potential negative consequences of using the following prompt in a classroom?
“Make a presentation on current events in your country.”
ChatGPT Answer
There are several potential negative consequences of using the prompt “Make a presentation on current events in your country” in a classroom setting:
Bias: Students may only focus on certain events or perspectives, leading to a biased understanding of current events.
Limited perspective: Students may only present events happening in their immediate community, leading to a limited perspective on current events in the country as a whole.
Inappropriate content: Students may present on sensitive or controversial topics that may be inappropriate for a classroom setting.
Lack of critical thinking: By only presenting current events, students may not develop the critical thinking skills needed to analyze and understand the underlying causes and consequences of the events.
Limited access to information: Some students may not have access to current events information, especially if they come from low-income or marginalized communities.
It’s important to consider these potential negative consequences and take steps to mitigate them, such as providing guidance on how to present information in a balanced and fair manner and encouraging critical thinking and analysis of current events.
Watching for potential bias
During my “conversation” with the bot, I was impressed with the response to the diversity and inclusion question I posed. The points it presented were, like most answers in this experiment, fairly common and expected responses for such a course (‘don’t talk about politics, religion, etc. etc.’). A phrase that kept cropping up is ‘it is important to…’ followed by a value statement. Those presented in this conversation were benign, and actually good advice from an educational standpoint (e.g. every student is unique, you should create a safe learning environment, and so on). It seems a lot of work has been done to minimize bias – responses from the bot on many issues have disclaimers and things that are ‘important to remember’. It is also, however, important to remember that biases can creep in and to keep an eye out for it, especially when taking advice from AI.
Copyright Issues
Obviously the chatbot is pulling this information from somewhere (more info on this is contained on the Wikipedia page for ChatGPT), and even though it is generative AI, one has to wonder how much this will be an issue in the future if content is too similar to an existing program or course – not to mention all other subjects it is capable of generating.
Five years ago, I wrote a small entry in the ORTESOL Newsletter about the then state of “adaptive software capable of teaching, testing, giving feedback, and most importantly, adjusting to student needs” (Chambers, 2015: 13). I mentioned a set of technologies colloquially referred to as the “Digital Aristotle”, or ‘Project Halo’ (Friedland et al., 2004), and the update to this six years later (Gunning et al., 2010). The Digital Aristotle was described as ‘an application that will encompass much of the world’s scientific knowledge and be capable of applying sophisticated problem-solving to answer novel questions’ (Friedland et al., 2004).
At the time, I was writing about a more grandiose piece of software that might one day replace the repetitive tasks of an ESOL classroom. The idea, or perhaps the concern about this technology for existing teachers was where they would fit in once a set of algorithms could replicate much of the day to day learning of a language course.
Five years on, I turn to how learning designers might be able to incorporate AI into course design.
Potential
First, a question: is there currently a program capable of teaching a course and adapting to student needs like an instructor? The answer is still no. Certain technology has, however, progressed to the point that portions of an online course can be enhanced by AI.
Adaptive Learning
Quite possibly the most exciting development in artificial intelligence for learning is that of adaptive learning. This concept has been in the works for a while with certain platforms utilizing algorithms to produce content that adapt to specific learner needs.
Instructure’s Canvas for instance, allows Instructional Designers to set up ‘Differentiated Assignments’ (Canvas Doc Team, How do I view differentiated assignments with different due dates in a course?, 2017) and ‘MasteryPaths’ (Canvas Doc Team, How do I use MasteryPaths in course modules?, 2020) which ‘allows targeted learning activities to be assigned to different users and sections’ (‘MasteryPaths’). Currently this is a manual process with course designers creating every assessment beforehand. The best students might not see the additional activities. It is not ‘intelligent’ in the way that course content is adapted specifically for a learner’s needs and on the fly. To do this requires large amounts of data and most importantly, AI training to see patterns, strengths, and weaknesses for a particular learner.
This is why Duolingo records progress at every step and offers learners a review of concepts the learner struggles with in previous activities. Platforms that provide an automated review often use large question banks and flag questions that learners initially, or continuously incorrectly answer. An intelligent AI could create novel questions based on learner goals, data from prior students, and information about the subject matter. An example of this is Google or Amazon’s ability to predict and offer products or suggestions based on the vast amount of information provided to them every single day. Certain training providers are currently working with application developers to produce tools capable of this on-the-fly feedback and adaptation.
Discussions
At the OLC Innovate 2020 conference, Kasey Gandham from Ed Tech company Paperback and Kim A. Scalzo, Executive Director of Open SUN, demonstrated how Paperback’s AI is being used with online discussions to help students write higher quality posts. As students write their discussion posts, the AI program checks for “close-ended questions, plagiarism, insufficient length, content about class logistics, profanity and abuse’ (Gandham & Scalzo, 2020). After this, if required, the post is moderated and the learner receives email feedback saying why and how to revise their post. The AI is also capable of suggesting posts to feature as the best of the week by analyzing, among other things, sentence depth and ‘curiosity score’.
More than a Quiz
The role and importance of ‘big data’ in online learning cannot be understated. Technology already exists that records the time learners interact with learning materials. It knows where they are clicking/tapping on the screen and how long activities hold learner attention. Using this data, AI could suggest, or even craft assessments that are adapted specifically to a single learner’s usage habits. Traditional quizzes which assess information retention could become only part of the larger formative assessment of the entire course, at every point in the course, without the learner even realizing any of this is happening.
Future
AI-enhanced design has the ability to transform Instructional Designers like never before. It could help us to modify our own design practices based on how learners are responding to course content. Through learner feedback, it could demonstrate which activities are most appealing and conducive to personalized learning goals. Big data’s role in recording learner interactions with content can provide insights into preferred learning styles and methods of instruction. Instructional Designers will have to continue adapting with the technology just as we have done in our everyday lives.
Chambers, P. (2015, Spring). “Digital Aristotle” and ESL: What does it mean for us?. ORTESOL Quarterly Newsletter, 38(1), P.12-13.
Friedland, N.S. et al. (2004). ‘Project Halo: Towards a Digital Aristotle’, American Association for Artificial Intelligence, 25(4), pp. 29-47. DOI: http:// dx.doi.org/10.1609/aimag.v25i4.1783.
Gunning, D. et al. (2010). ‘Project Halo Update – Progress Toward Digital Aristotle’, Association for the Advancement of Artificial Intelligence, 31(3), pp. 33-58. DOI: http:// dx.doi.org/10.1609/aimag.v31i3.2302.
