Since the launch of ChatGPT on November 22nd, 2022, a proliferation of innovative artificial intelligence (AI) tools has swept through the digital landscape, each promising advantages in speed, functionality, and performance. While this rapid adoption is exciting, users are struggling to keep up. This article explores the practical application of these tools in higher education. The technology presented here has some complex underpinnings, but no advanced technical knowledge is required to understand or apply the examples that follow.
rapid engineering or directing technology
The main difference between average and advanced users of AI models is not technical skill, but their ability to give clear instructions. In the AI world, this skill is called prompt engineering. A great way to explore this skill is by looking at his MidJourney, the most popular image generator.
MidJourney is a leading AI image generator, so its general output defaults to higher quality than most of its competitors, but rapid engineering still makes a big difference.
Ask MidJourney to generate a portrait of a graduate student you need to use in your photography project. In this case, the prompt given to MidJourney would be ‘Portrait of a graduate student’ and, as expected, the results are pretty good, but it may require some extra work to meet the project’s requirements.
If you put on your quick engineer’s hat, you can augment this request with very specific instructions to MidJourney to ensure the output is as close to your needs as possible. For example, MidJourney can understand camera models, lens models, manual camera settings, and most importantly context. The more information you enter, the more likely you are to get exactly what you’re looking for. So let’s add some more instructions to the prompt. “Portrait of happy graduate student in gown and hat on campus. Shot with Canon EF 24-70mm f/2.8L II USM –ar 16:9.”
Well this is what you got. Note that by default MidJourney only creates square images, so in this example he also specified a 16:9 ratio.
Chatbot Assistant Prompt Insertion and Creation
Don’t let the title of this section scare you. I promise it’s not as technical as it sounds. Injecting prompts in this case means giving the chatbot a specific set of instructions so that it can consider each response.
This is useful, for example, if you want to stay on a particular topic without having to repeatedly tell the chatbot every time you interact. Poe is a free service created by his popular Q&A platform Quora that allows anyone to create a personalized chatbot using his OpenAI or Anthropic AI engines.
Sign up for free and click Create Bot. Create a travel assistant and give clear instructions. You can give your bot the name “GlobeTurtler”, a description, an introductory message, and most importantly insert a prompt in the “Prompts” section. This is the most important step as all interactions with this bot are based on it. In our case we have a very specific use case.
You are a trip advisor providing travel advice to vegetarian British citizens. Provide the following information in case of questions:
- visa requirements
- The easiest way to travel between specified countries or cities
- Recommended food
- Top 3 tourist attractions
- Top 3 tips for travel safety
Now you can enter your preferred country and city into “GlobeTurtler” and you will always get the desired output.
Exact processes can be replicated for virtually any topic or area, saving time and eliminating the need to repeat prompts when answers are needed. The instructor instructs the bot to act as a classroom assistant in a particular field (such as a modern diplomacy course) and respond to a given topic with exercises suitable for a group of up to 6 students suitable for that course. may ask you to do so. Certain skills such as teamwork and debate are required.
Remember: All current models are error-prone and can sometimes display incorrect information, so always double-check the output before running.
Using artificial intelligence in research
Chatbots are great assistants for generating answers to texts and questions, but other models are even better when it comes to summarizing and speeding up surveys. WordTune Read is a service that provides full summaries of articles and books of up to 300 pages. A free version is available for up to 3 articles per month, making content more quickly available to scholars and students.
The features that distinguish WordTune from other tools are the ability to cross-check summaries for correctness in real-time and to re-summarize specified paragraphs. As a test case, we gave him his 2022 report of the United Nations on the Sustainable Development Goals (SDGs). The left pane provides a summary of the report, while the right has automatic synchronization with the source of the summarized paragraphs for real-time fact-checking.
Another unique feature of this tool is semantic search Options allow you to summarize documents based on keywords. We asked WordTune to summarize the UN report based on the keyword ‘education’, omitting the remaining information. This can make a big difference when looking for very specific information in large texts. For example, it may take at least several hours to extract all educational information from a manual, but only a few minutes using the semantic search feature.
put it all together
We’ve seen quite a few use cases for AI technology, but there’s another great service that goes beyond quick insertion and summarization to help create chatbots with the help of knowledge bases. An advantage of using a knowledge base is that it reduces the chances of chatbots providing incorrect information. MeetCody allows you to create chatbots that can only answer questions based on a configured knowledge base. MeetCody is free to start and allows you to try up to 100 interactions with your bot.
To get started, sign up and add a knowledge base to use for training your bot. A knowledge base can consist of Word documents, PDFs, PowerPoint files (up to 100 MB each), or links to websites containing text. You can give your bot a name, description, company affiliation, and detailed instructions on what to do, especially if you don’t know the answer. It can be very strict to only look at the knowledge base, or it can be set to fill any gaps (although it may contain inaccurate information).
A nice feature is that it can be trained to serve a URL of your choice when the answer to a user’s query is not found.
So we uploaded the Nottingham Trent University public international partnership documents to our knowledge base and configured our chatbot to be very strict, referencing only these documents for answers. If they don’t have an answer, I asked them to direct the user to a relevant link of my choice.
Once you’ve set up and tested your bot, you can share it using a link or embed code on your website.
With this tool, you can create highly specialized chatbots that help internal and external users understand university processes and policies, eliminating the need for human input when answering simple questions. Reduced.
Kato Lorea is Senior Global Partnership Manager and Technical Consultant at Nottingham Trent University.
If you want advice and insights from academics and university staff delivered directly to your inbox each week, Sign up for campus newsletter.
