Why Artificial Intelligence matters to us all


Lynn's Cat produced by Dall-E

Lynn's cat as a writer produced by OpenAI's DALL-E

By Associate Professor Lynn Gribble, UNSW Business School

Published 4 April 2024


Open AI and its Large Language Model (LLM), better known as ChatGPT, celebrated its first birthday in November 2023. Today, it is time to recognise Generative AI (GenAI) for what it is: a tool that is freely available, that will be used by our students in their future workplaces, and for the assistance it can provide to make our lives more efficient and generally easier. Rather than fear or ignore it, the task at hand is to integrate GenAI into our daily lives as we have done with search engines, word processors and other technology including smart phones.

As GenAI is still considered relatively new technology, it can be helpful to look at where you are in your own journey of adoption. Rogers (1962, 2003) seminally notes, in the diffusion of innovation there are several categories based upon how people respond. From the innovators (some call them evangelists), early adopters, early majority, late majority, to the laggards, all adoption types coexist as the technology continues to develop. While the innovators and evangelists have enthused over the capabilities and its outputs, the early adopters are now also using it regularly in their work. For the early majority, they are still coming to terms with what it might mean for their practice, and meanwhile the late majority and laggards are hoping someone, anyone, will just tell them what to do, or if they ignore it for long enough it will go away.

Today, some 16 months after GenAI became freely available, it is time to consider who and where we are, because this enables us to be empowered in our own journeys and take the next steps in the adoption of AI which is now both ubiquitous and freely available. It is important to consider that GenAI is a tool and like all tools, it can make life easier; and the tool itself is neither inherently bad nor good, however, we must acknowledge that its use might be.


A recap for clarity

In considering what might be the next steps in any AI adoption journey a ‘state of the nation’ recap can assist in setting the scene. OpenAI owns none of the information it uses. It collects (note, it does not search, as an LLM operates on what has been collected) information which is (often) out of date, inherently biased, and produces a programmed output that is designed to be creative (its tolerance for accuracy is naturally set at 0.7 ‘temperature’ with temperature being the parameter of operation) meaning it is likely to make up both information and citations. It has also been programmed to provide ‘plausible’ output meaning the output can appear accurate even if it is not. The output, which is often bland, generic, and lacking any depth, is yet another problem to be considered. While some of the GenAI tools for research are less problematic, fact checking remains an imperative. The continual need for a human to take responsibility for the output and information provided is required with any use of GenAI along with knowing when and how to acknowledge its use.


The current landscape

While some may continue to resist the use of AI, those who have adopted it have found significant productivity gains. To determine when and how to use GenAI, a specific exploration of the academic workplace and classroom is helpful.

One way the academic workplace can be characterised is by administration, teaching, research, and engagement. While the physical acts of teaching and engagement are best undertaken by a human, the administrative and research acts and, in turn, the load created, can be ‘lightened’ by ‘employing and deploying’ some GenAI. To do this, it is helpful to reimagine the administrative load. What tasks could another ‘person’ do? Then play with GenAI to do this work. From drafting communications, to writing outlines, or step-by-step guides, rubrics and more, GenAI is a helpful assistant that needs close supervision.

There are also many tools that can be helpful in the research process too, such as Research Rabbit, Scopus AI (with the highest access to subscription journals) and Elicit (open access only). They can all conduct basic literature searches and provide some collated outputs (in an annotated bibliography style) drawing together the largest themes or summarise information you have given it (remember its limitation is working only with collected information it currently has stored as a seeking over searching tool).

GenAI can take any piece of information and summarise it or make suggestions based on the content ‘seeking’ what you have asked it to find. Be careful however, that you might be feeding the ‘beast’ and a word of warning, UNSW does not permit the loading of student assignments/assessment into any AI at present. However, you could ask it to phrase a comment about a consistent problem you have observed and how this could be addressed and then use that in your feedback.

The possibilities with GenAI are endless. However, if you commence by considering why are you doing something and ask yourself if a machine could do it more efficiently, then a ‘GenAI’ might become your new ‘best friend’. Just keep in mind the need to supervise it, edit and rewrite what it produces so it maintains your authentic voice, as you might if you partnered with another person.

In the classroom, considering how our students may use AI in their future workplaces can improve AI literacy. Deployed in this manner, GenAI use in the classroom can support authentic learning. It can be an excellent tutor, and well deployed, can support students 24/7 to find information or answer questions. It can also assist with the drafting, editing or critical review, if that is how you want your students to use it.

Alternatively, consider how they are or might be using it and set some solid expectations about its use, along with altering assessments and grading criteria to include the acknowledgement of GenAI along with how it might be impacting what is being measured. (See UNSW’s AI referencing guide for students here)


The question remains of where to start

Getting started with any new technology can feel overwhelming, and I observe the sharing of prompts with amusement. Each time I am asked to share prompts, I refuse. The reason, quite simply, is learning how to prompt is a key skill for the future and taking someone else’s prompts won’t develop this skill and can lead to a bigger well-known problem of any technology output which is often referred to as garbage in/garbage out.  Learning prompt conditioning is more useful.

Prompts need to be developed in consideration of the role you want GenAI to play or take. Is it a tutor, an expert or a reviewer, or something else? Defining the task is also important along with the context, that is what would the AI do to create the output and what are the conditions of its context. And finally, using AI means engaging in an iterative process. No one is a ‘one prompt’ wonder! Refine, refine, and refine. Set aside time to ‘play’. If you see a good prompt from someone, examine it. Consider why you think it is good, what you would want to change, and why the next use of the prompt might result in a different answer (remember it’s seeking not searching!). Prompts are not a standard maths equation that you can rote learn and use, they require refinement over time.


Some observations

At this stage in the diffusion of innovation, every (current and potential) user needs to understand what it can do and how it does it, as well as how the technology is evolving. Broad ethical considerations of what it means to use such tools also takes time, with many publishing houses still banning or limiting its use. Australia has released some ethical consideration for the use and development of GenAI and the EU is further along in its considerations, but this is still considered an emerging field. This cannot be forgotten at this stage.

Using GenAI has reduced the time I spend on administrative and support tasks, freeing up cognitive load to engage in more creative and considered actions that have broader-reaching impact. GenAI is my muse, idea generator, editor, reviewer, and administrative assistant but it is not my partner. While I always treat it as a person, it needs constant supervision. Yet by giving it a persona I tend to consider more deeply my expectations and the outputs.

From a student perspective, they are learning too. Much of the work submitted as a result of GenAI is often bland, incorrect, or contains inaccurate information. Lack of content knowledge leads to acceptance (unquestioned) of what it has produced. Adopting the tool without understanding its limitations and the ethical considerations with its use leads to other problems in the resultant work, too. 

It is our role, as educators, to engage our students in the ethical use and exploration of GenAI to prepare them for the future. Further, developing student AI literacy is a must. The future means both we and our students need to capitalise and exploit these tools as they are commonplace. We are surrounded by GenAI, our job now is to employ it.


Looking for AI in higher education teaching resources?
  • See what other UNSW academics say about AI in education
  • UNSW colleagues can access resources about GenAI in teaching and assessment here 
  • UNSW’s AI referencing guide for students can be found here


Read other UNSW Education blogs or submit your own


Reading this on mobile? Scroll down to learn about the author.


Enjoyed this article? Share it with your network!