Should Instructional Designers use Chat GPT?
- What is Chat GPT?
- Some other popular GPT-2 Tools
- What sort of knowledge is needed to utilize Chat GPT effectively?
- Creative thinking and strategy
- How are people currently using AI chat tools in education and training?
- What are the limitations of using GPT-3?
- What is the best way to integrate Chat GPT into the workflow of an instructional designer?
- What role is GPT-3 playing in the future of eLearning
Before getting into the virtues of AI and Chat GPT, let’s establish a few things.
As we have said in a previous article, there are many ways that artificial intelligence can make your job easier. But, like anything that runs on inputs to generate outputs… rubbish in, rubbish out. If you don’t have the fundamental domain knowledge then no amount of robots can help you build anything of value.
“Armed with prompts which are carefully crafted to ask the right thing in the right way, educators can use AI like GPT3 to improve the effectiveness of their instructional practices.”
If your inputs are well constructed (because you have a high level of domain knowledge) AI can make a good course better.
So, can AI tools and specifically Chat GPT help educators and instructional designers?
The short answer is “Yes”.
But first, let’s look into what exactly Chat GPT is and why it’s so popular.
Did you know a learning designer’s average salary is 70k?
What is Chat GPT?
Chat GPT is short for Chat Generative Pre-Trained Transformer. This is an advanced chatbot version commonly referred to as ChatGPT.
It was created by OpenAI in November 2022 and had one of the fastest adoption rates of any technology in history.
The quality of the output is so high that there have been various reports of ChatGPT passing a Turing Test but not in an official capacity, like the Loebner Prize.
This is the third iteration of ChatGPT, the first launched in 2018.
GPT-3 now uses longer lexical analysis, and a much wider set of parameters (175 billion) and combined with the advancements in natural language processing (NLP) it means that the pre-training element has become so refined it has eliminated the need for human supervision.
Shortly after ChatGPT was unveiled by OpenAI in September 2020, Microsoft announced it had agreed to “exclusive” use of the GPT-3 model in ChatGPT.
While we can still use the public API, only Microsoft has access to GPT-3’s underlying model. It may be a matter of time before it becomes a completely paid-for service.
Some other popular GPT-3 tools
ChatGPT is certainly not the only AI chatbot tool in the market. Here is a list of some of the other popular alternatives.
- OpenAI API: The creators of ChatGPT have a platform that provides developers and businesses with access to GPT-3 capabilities, enabling them to create custom AI-powered applications.
- Replika: A personal AI-powered chatbot that can be used for self-reflection, therapy, and mental wellness.
- Jarvis.ai: A content creation tool that utilizes GPT-3 to generate articles, blog posts, product descriptions, and other types of content in a matter of minutes.
- Lobster: A writing assistant that uses GPT-3 to help users generate ideas, write headlines, and improve the overall quality of their content.
- Muume: An AI-powered project management tool that uses GPT-3 to help teams organize their work and improve productivity.
- Infiny: A conversational AI tool for e-commerce that enables businesses to provide customers with instant access to product information, recommendations, and support.
- Typeracer: A typing game that uses GPT-3 to generate texts for players to type, helping users improve their typing speed and accuracy.
Like with innovations, there will be more and more applications of GPT-3 and ChatGPT clones emerging in the coming month and years. It’s also highly likely that search engines will begin to update their algorithms to detect AI-generated content. There are already programs that can scan content to see if it was written by a human or a machine.
What sort of knowledge is needed to utilize Chat GPT effectively?
As previously mentioned, AI tools are not a substitute for domain knowledge. In reality, the knowledge you have on learning design is more helpful than the knowledge you need for working with ChatGPT. For example, to utilize Chat GPT effectively, an instructional designer might benefit from knowing the following:
- Knowledge of AI and natural language processing (NLP) is crucial for effectively using Chat GPT, as it involves working with machine-generated text and dialogue.
- Familiarity with the API of the Chat GPT platform being used, including how to input and receive data, and how to customize and control the output.
- Technical skills, such as coding, may be necessary for integrating Chat GPT into e-learning platforms, setting up data flows, and customizing the outputs to meet specific design needs.
However, it is far more important to have a really good understanding of learning and/or instructional design principles and a deep understanding of the course content. This means you will be able to quality control the output generated by Chat GPT so it aligns with the instructional goals and accurately reflects the intended learning outcomes.
Rather than learning ChatGPT inside out, a better use of time would be in preparing and planning your course and scoping out the materials the learners will require.
Creative thinking and strategy
As an instructional designer, utilizing Chat GPT effectively requires a combination of creative thinking and strategy to produce the desired output.
Here are some key considerations for each input to achieve the desired output with Chat GPT:
1 – Inputs:
The inputs to Chat GPT should be well thought out and reflect the instructional goals and desired outcomes. The input should include clear and concise questions or prompts that guide the AI in generating the desired output.
2 – Creative thinking:
Creative thinking is important in developing the inputs, as it allows the instructional designer to think outside the box and consider different approaches to the problem. For example, instead of asking straightforward questions, the instructional designer could present scenario-based prompts to generate more engaging and interactive outputs.
3 – Strategy:
A clear strategy is necessary to guide the input and ensure that the output aligns with the instructional goals. The strategy should include a clear understanding of the course content and desired learning outcomes, as well as a plan for verifying and refining the output generated by Chat GPT.
4 – Iteration:
Iteration and refinement are important for ensuring the desired output. The instructional designer should review the output generated by Chat GPT, make adjustments as needed, and repeat the process until the desired outcome is achieved.
It’s all about the Prompts
The key to success with GPT-3 is in the writing of the prompts. Prompts are the instructions you type into ChatGPT for it to create the text. Naturally, the better the prompt, the better the output.
Here are some helpful tips for writing prompts for ChatGPT with learning design in mind.
Identify the learning objectives:
Before writing prompts, it is important to have a clear understanding of the learning objectives for the course materials. This will help you quality control the outputs you need from GPT-3 and ensure that the prompts you write align with these objectives. Even the slightest changes to the words in the prompt can drastically change the quality of the outputs.
Define the structure:
Decide on the structure of your course materials. For example, will it be a lecture-style presentation, a video script, or a storyboard? This will help you determine the type of outputs you need from GPT-3 and how to format the prompts you write. Give clear direction in your prompt – Write a list, write a short paragraph, and give examples of 5 headlines.
Write the prompts:
Write a series of prompts that will guide GPT-3 to generate the outputs you need for your course materials. Be specific (see above points) about the type of information you want to include in each output, and use language that is easy for GPT-3 to understand. Don’t settle for the first answer if you’re not happy.
Use examples to help GPT-3 understand the type of outputs you want to generate. This can include examples of previous course materials, content from other sources, or your writing.
Test and refine:
After writing your prompts, test them by inputting them into GPT-3 and reviewing the outputs. Refine your prompts as needed to ensure that they generate the outputs you need for your course materials.
Once you are satisfied with the outputs generated by GPT-3, integrate these outputs into your course materials. This may involve editing and revising the outputs to meet your needs or using the outputs as-is.
Once you’ve spent some time working with ChatGPT you appreciate it’s not as simple as prompt, copy and paste. Overall, writing prompts for GPT-3 is an iterative process that may take some time and experimentation to get right.
However, with a clear understanding of the learning objectives and a focus on writing specific, clear prompts, it is possible to create high-quality course materials.
How are people currently using AI chat tools in education and training?
People are currently using AI chat tools in education and training to create more personalized and interactive learning experiences. Here are some good examples of how AI chat tools are being used in training:
- Virtual tutors:
AI chatbots can be used as virtual tutors to provide students with personalized assistance and feedback on their coursework.
- Personalized learning:
AI chatbots can be used to create customized and personalized learning experiences, such as tailored reading recommendations and self-paced quizzes.
- Language learning:
AI chatbots can be used to provide conversational practice for language learners, allowing them to practice their speaking and listening skills.
- Student engagement:
AI chatbots can be used to increase student engagement and motivation through interactive quizzes, games, and discussions.
- Remote learning:
AI chatbots can assist with remote learning, providing students with instant access to information and resources, and helping to manage and organize course content.
What are the limitations of using GPT-3?
As we mentioned at the outset, AI, machine learning, chatbots and ChatGPT are not without their limitations. Like baking a cake, the output is only as good as the ingredients and the receipe.
The ChatGPT home screen lists some of the limitations of the tool, specifically that it
- May get things wrong
- May give harmful instructions
- Is not too sure about the world after 2021
If we consider all that’s happened in the brief period since 2021 you begin to understand why ChatGPT is not a silver bullet.
The limitations of using GPT-3 specifically in digital learning design include
1 – Bias and lack of diversity:
GPT-3 is trained on vast amounts of text data from the internet, which may contain biases and lack diversity. This can result in output that reinforces existing stereotypes and biases, and may not be representative of diverse perspectives and experiences.
2 – Lack of control over output:
While GPT-3 is incredibly versatile and can generate a wide range of outputs, instructional designers may not have complete control over the content generated by the tool. This can be problematic when trying to ensure that content is aligned with specific learning objectives and instructional goals.
3 – Quality of output:
GPT-3 is capable of generating high-quality outputs, but the quality of the output depends on the quality and specificity of the input. If the input is vague or poorly defined, the output may not be as useful or relevant.
4 – Ethical considerations:
The use of AI-generated content raises ethical considerations, such as the potential to perpetuate misinformation and the need to ensure that AI-generated content is differentiated from content created by humans.
5 – Dependence on external data:
GPT-3 is trained on vast amounts of external data, which can limit its ability to generate outputs based on more specialized or niche subject areas.
Once these limitations are understood, GPT-3 has the potential to be a powerful tool for instructional designers.
Become certified in instructional design with a professional diploma in digital learning design.
What is the best way to integrate Chat GPT into the workflow of an Instructional Designer?
The best way to integrate Chat GPT into the workflow of an instructional designer is to focus on the areas in which it can be of most assistance. With good clear prompts, Chat GPT can be valuable. If you think it will replace the hours or preparation and years of domain knowledge this is where you will run into trouble.
The actual application will vary depending on the specific needs and goals of each project. However, here are some general steps that instructional designers can follow to effectively integrate Chat GPT into their work:
Define the project goals:
Before using Chat GPT, it’s important to have a clear understanding of the project goals and desired outcomes. This will help guide the input and output of the Chat GPT model.
Prepare the data:
Instructional designers will need to prepare the data and content that will be used to train the Chat GPT model. This may involve sourcing existing materials or creating new content.
Train the model:
Next, the instructional designer will need to train the Chat GPT model using the prepared data. This process will involve fine-tuning the model to meet the specific needs of the project.
Test the model:
Once the model has been trained, the instructional designer should test it to ensure that it is providing the desired output. This may involve conducting a pilot study or testing the model with a small sample of users.
Incorporate into the course:
Finally, the instructional designer should integrate the Chat GPT model into the course or e-learning platform. This may involve creating interactive quizzes, virtual tutors, or other engaging learning experiences.
By following these steps, instructional designers can effectively integrate Chat GPT into their workflow and create engaging and personalized learning experiences for students.
What role is GPT-3 playing in the future of eLearning
The future of eLearning and AI-powered tools in digital learning design is bright. As technology continues to evolve and play a larger role in education, we will see new and innovative applications of AI will emerge in the field of eLearning. We may only be seeing the beginnings of how this technology could transform the way we approach digital learning design.
Here are a few trends that are shaping the future of eLearning at the moment:
AI-powered tools, such as Chat GPT, will become increasingly sophisticated, allowing for more personalized learning experiences based on individual student needs and preferences.
Gamification and other interactive elements will become increasingly prevalent in eLearning, making learning more engaging and motivating for students.
- Integration with other technologies:
AI-powered tools will become more integrated with other technologies, such as virtual assistants and wearable devices, to provide a more seamless and comprehensive learning experience.
- Virtual reality and augmented reality:
Virtual and augmented reality technologies will become more widely adopted in eLearning, providing students with immersive and interactive learning experiences.
- Increased accessibility:
AI-powered tools will play a key role in making education more accessible and inclusive, particularly for students with disabilities or those in remote or underserved areas.
- Content creation:
GPT-3 can be used to generate high-quality course materials, such as lecture notes, summaries, and quizzes, at scale and in a fraction of the time compared to manual content creation.
- Assessment and feedback:
GPT-3 can be used to provide real-time feedback to students on their progress and understanding of the material, helping to identify areas where they need additional support and resources.
GPT-3 can be used to create interactive learning experiences, such as virtual tutors, that simulate human-like conversations and respond to student questions and feedback in real-time.
Overall, GPT-3 has the potential to significantly improve the quality and efficiency of eLearning, making it more personalized, interactive, and accessible for students. As technology continues to evolve and AI models like GPT-3 become more sophisticated, it’s likely that we will see even greater innovations in the field of eLearning, further transforming the way we approach digital learning design.