In late 2023, OpenAI released the hotly anticipated new feature of GPTs to the world. This development could be a major leap forward in making AI more accessible and personalised. Read on to find out more about GPTs and their applications.
By Rick Grammatica and Denis Mallon
What are GPTs and how can you build one?
The latest feature introduced by OpenAI allows users to create custom versions of ChatGPT tailored for specific purposes, known as GPTs.
Anyone with a ChatGPT Plus account can build their own custom GPT. The GPT Builder is an intuitive user interface, allowing users to define the purpose of their GPT, set parameters for what it should and shouldn't do and upload training documents.
If you have a ChatGPT Plus account, select ‘Explore’ from the sidebar and then ‘Create a GPT’. There are two ways of building the GPT, Create and Configure, as well as a preview where you can test your chatbot while you build it.
The Create feature allows you to work with the AI to build the GPT collaboratively.
This is useful if you’re open to ideas or are unsure how to write the instructions yourself.
The Configure option requires you to enter the details manually, giving you more control over the instructions and knowledge used to customise it.
Writing the instructions is like writing a detailed prompt or custom instruction.
After configuring your GPT, it can be made public, kept private or shared via a link. In the next few weeks, according to OpenAI, you'll be able to sell your GPT in a GPT store.
Why are GPTs an important development?
According to the Nielsen Norman Group, AI systems will likely create a new way of interacting with technology. Instead of giving step-by-step commands, users express what they want to achieve and the AI works out how to do it. An example of this would be an autonomous AI agent that performs longer or complex tasks from just a few simple instructions.
GPTs are a step in this direction, as they can be designed to help us perform certain tasks more effectively and efficiently. GPTs can be linked with third-party plugins, such as Zapier, to perform simple actions in a variety of apps, like Outlook. This type of interaction can be beneficial in learning environments, where clear and simple interfaces make it easier for the learner to focus on learning and reach their goals.
Personalised chatbots aren't new and have been used in customer service, education and other sectors for years. However, GPTs are making these more accessible, enabling a wider range of people to create customised chatbots without needing advanced technical knowledge. With GPTs, you can program the AI to your specific needs using simple, everyday language — as long as you give enough appropriate context and instructions.
How can GPTs be used for learning?
GPTs can be personalised for specific needs or tasks, so have a range of applications to learning and learning design. Here are a few examples.
Guided learning experiences
The GPT can be trained with project-specific resources and information about the learner and teaching methods. For example, high school students could complete a project about the Second World War with the help of a customised GPT.
Subject matter expert
A GPT can serve as a virtual subject matter expert. Imagine a GPT for students, teachers or designers trained on a particular topic, capable of answering intricate questions, presenting case studies or even debating current issues in the field.
Learning designer
For educators and learning designers, a GPT could be a valuable tool in creating and refining educational content. It could assist in developing course outlines, lesson plans, activities and assessments. This could particularly benefit those less familiar with instruction design approaches.
Training scenario simulator
In the Learning & Development sector, a GPT could be used to create realistic training scenarios. For instance, in a sales training program, a GPT could simulate customer interactions, allowing employees to practise their communication and problem-solving skills in a controlled, yet dynamic environment.
Research assistant
For Higher Education, a GPT could serve as a research assistant, helping students and educators gather and summarise academic papers, create bibliographies or even draft initial research proposals. This application could significantly streamline the research process, allowing more time for in-depth analysis and discussion.
Language coach
A "language coach" GPT could offer personalised language learning assistance. It could simulate conversation scenarios, provide grammar coaching and cultural insights, making language learning more interactive and accessible.
What next for GPTs?
GPTs mark a step towards more personalised, intuitive and efficient tools for learning designers, teachers and learners. If GPTs explode in popularity, there’s the potential of the GPT Store to be the new Apple App store, where users have access to an array of customised chatbots and can make money by selling their GPTs. As AI technologies develop, we may see autonomous agents in the near future. These personalised assistants would perform tasks for you independently, like researching and writing an essay or presentation.
However, GPTs aren’t perfect. There remains a need for domain-specific knowledge to effectively build GPTs. First, knowing how to write effective instructions will help you build the GPT. Second, while ChatGPT has a broad understanding of many topics, user input is needed to refine and direct this knowledge. So when designing GPTs for education, you should provide guidance about the topic and how to interact with the learner.
Security concerns have also been reported, where GPTs have been prompted to provide the custom instructions and files used to build it. Furthermore, there remains the problem that ChatGPT can provide incorrect information, so even if a GPT is customised to play a certain role, there’s no guarantee it’ll be correct.
There’s still a way to go before GPTs are fully secure, accurate and accessible to a wider audience. Nevertheless, they are a step forward towards a more intuitive and customised experience for the user and learner.
Want to learn more?
Follow Rick Grammatica on LinkedIn and subscribe to this blog by selecting the button in the header or footer. Check out other blog posts for more ways AI can be used for learning design.
Follow Denis Mallon for his thoughts about helping humans work smarter, through design and technology.
Watch this short tutorial about how to use GPT Chat Builder.
This Nielsen Norman Group article, ‘AI: First New UI Paradigm in 60 Years’, discusses how AI is introducing the third user-interface paradigm in computing history.
Daniel Miessler writes about AI security risks, including those relating to GPTs, on LinkedIn and his blog Unsupervised Learning.
Want help building a customised GPT? Or to streamline your learning design processes? Contact me on LinkedIn.
Comments