- Introduction
Background
1.1Generative artificial intelligence (GenAI) is rapidly affecting the global technology landscape and changing the way we learn, work, and interact with each other. The use of GenAI as a tool in education is a recent development that is not widely understood. It comes on the back of the COVID-19 pandemic which introduced more educational technology (EdTech) into our homes as part of remote education systems.
1.2The introduction of GenAI in education received a mixed response globally and within Australia. Some countries like Italy initially banned the use of ChatGPT due to privacy concerns. Others, like China, encouraged the use of GenAI, but regulated it early. Some Australian jurisdictions banned its use in the school system, and some universities also prohibited GenAI while others embraced it.
1.3Now, the Australian Government has a framework to encourage schools to use GenAI, and more universities are also integrating GenAI. There is a growing acceptance that GenAI is here to stay and that Australia needs to evolve with it. Thistrend is being met with increased support for GenAI’s integration into the education system, and for guidance and safeguards to assist students, staff, parents and guardians, and developers.
1.4The potential for GenAI to deliver benefits in the Australian education system, like personalised learning, improved education outcomes, and streamlined administrative processes, is exciting. Worldwide, there is considerable interest in learning how to maximise these opportunities.
1.5However, there are also significant challenges relating to the use of GenAI, including broader safety and security concerns around ethics, data, and privacy. Theparticular vulnerability of children heightens many of these risks. The absence of a strong evidence base regarding GenAI’s impacts in education, despite the surging availability and use of the technology, is also an underlying concern.
1.6With the rapid uptake of GenAI by students and educators domestically and internationally, and the benefits and risks surrounding that, it is essential to consider how to best integrate GenAI into Australian education and manage its use.
GenAI tools
1.7Leading artificial intelligence (AI) developers are reaching masses of consumers via their AI systems, including OpenAI (Microsoft is its largest shareholder), Google, Meta, and startups such as Anthropic, Cohere, and France's Mistral. The use of GenAI soared with ChatGPT. On 30 November 2022, OpenAI made ChatGPT freely available to the public. ChatGPT has amassed over a billion hits worldwide per month since February 2023. Since the launch of ChatGPT, the competition to capture the market has intensified, and so too has the use of GenAI, including in the education system.
1.8On 16 January 2024, one year from the announcement that Microsoft would begin integrating GenAI into Azure, Microsoft launched Co-pilot Pro. The South Australian Department for Education (SA DFE) adopted this technology in its GenAI trial in public schools.
1.9Google released a GenAI chatbot called Bard on 6 February 2023. It was renamed Gemini in February 2024 to compete with OpenAI’s ChatGPT. On 28 November 2023, Amazon introduced a GenAI tool called Amazon Q. Meta released its most powerful AI model called Llama 3, and in April 2024, it released two smaller versions of that system and integrated them into the Meta AI assistant feature across Facebook, Instagram, and WhatsApp.
1.10GenAI is an economic game changer. Amazon Web Services (AWS) asserted that GenAI could produce $315 billion in economic value to Australia over the next decade. The Tech Council of Australia forecast that GenAI could contribute between $45–$115 billion in annual economic value to Australia by 2030. It predicted that benefits would be derived from the adoption of GenAI in existing industries, and from the development of AI products and services to create new jobs and businesses.
Understanding GenAI
1.11As the name suggests, GenAI generates content, including text, code, images, music, audio and video. GenAI is a subset of AI, and GenAI products are broadly situated within EdTech. Figure 1 represents the relationship between key concepts regarding GenAI:
Figure 1.1Relationship of key concepts
Source: ‘Glossary of AI terms, MinterEllisonRuddWatts, viewed 13 August 2024.
1.12AWS defines AI as:
the use of machine learning (ML) and related technologies that use data to train algorithms and predictive models for the purpose of enabling computer systems to perform tasks normally associated with human intelligence or perception, such as computer vision, natural language processing, and speech recognition.
1.13GenAI uses an emerging area of ML, called deep learning, to generate new content based on user prompts. The University of Sydney (UoS) explained that GenAI is:
arapidly evolving class of computer algorithms able to create digital content—including text, images, video, music and computer code. They work by deriving patterns from large sets of training data that become encoded into predictive mathematical models, a process commonly referred to as ‘learning’… People can then use interfaces like ChatGPT or MidJourney to input prompts—typically instructions in plain language— to make Gen-AI models produce new content.
1.14The very large ML models that drive GenAI are commonly referred to as foundation models. A specific type of foundation model is a large language model (LLM), which focuses on language tasks. The LLM is trained on large amounts of data scraped from the Internet, hence the data and outputs can be biased, inaccurate, and outdated. LLMs cannot identify whether information is factual and truthful. TheUniversity of Melbourne (UoM) explained that:
There is no inner logic, no reasoning, no repository of knowledge, and no explicit guidance to anything it does except the numerical match scores that provide segment continuation rankings. It simply recognises patterns in data and produces answers based on those patterns.
1.15LLMs sit behind and power GenAI tools that end-users employ, such as GenAI chatbots like ChatGPT. To illustrate, if you type a user prompt such as a question into ChatGPT, that prompt is processed by an LLM to produce a response.
1.16AI chatbots are being used in many sectors, including in education. Google described AI chatbots as:
…[A]pps or interfaces that can carry on human-like conversation using natural language understanding (NLU) or natural language processing (NLP) and machine learning (ML). AI chatbots differ from standard chatbots in that they leverage large language models (LLMs) versus traditional conversation flows and pre-programmed responses to generate responses to text and voice inputs.
Parliamentary inquiries
1.17There are other parliamentary inquiries underway on AI in Australia. In June 2023, the Parliament of New South Wales launched an inquiry into AI. That inquiry received some evidence about GenAI and education, for instance, from the Australian Education Union (AEU) and UoS.
1.18A Senate Select Committee on Adopting AI was also established on 26 March 2024 to inquire into the opportunities and impacts of the uptake of AI technologies in Australia. The Senate Committee intends to report to the Parliament in September 2024.
Referral and conduct of inquiry
1.19On 24 May 2023, the House Standing Committee on Employment, Education and Training (the Committee) adopted an inquiry into the use of GenAI in the Australian education system, following a referral from the Minister for Education, theHonJasonClare MP. The Committee was asked to focus on the issues and opportunities presented by GenAI and its impacts on Australia’s early childhood education, schools, and higher education (HE) sectors. The Terms of Reference can be found in the preliminaries part of this report.
1.20The Committee announced its inquiry in a media release on 25 May 2023, and called for written submissions. The Committee received over one hundred written submissions, which are listed in Appendix A.
1.21The Committee appointed a panel of AI subject matter experts to support the inquiry. Members of the expert panel included:
- Dr James Curran, Chief Executive Officer, Grok Academy
- Professor Nicholas Davis, Industry Professor of Emerging Technology and Co‑Director of the Human Technology Institute, University of Technology Sydney (UTS)
- Professor Leslie Loble AM, Industry Professor, UTS
- Associate Professor Julia Powles, Director of the Tech & Policy Lab, University of Western Australia (UWA).
- The purpose of the expert panel was to provide the Committee, at the outset of the inquiry, with an overview of key issues and trends regarding GenAI and education, and to help shape the collection of evidence. The panel acted as a sounding board and provided an opportunity for the Committee to engage with a group of experts who followed the inquiry and could comment on the breadth of evidence presented to it. The public hearing program was opened and closed with a meeting of the expert panel.
- The Committee held 15 public hearings virtually and in Canberra, Melbourne, and Sydney. The Committee visited Monash University (MU), Pymble Ladies’ College (PLC), and The Grange P–12 College. These site visits were undertaken as case studies of the breadth and degree of integration of GenAI in education. Transcripts for all public hearings are available on the Committee’s website, and details of the public hearings are listed in Appendix B.
- Most of the evidence received by the Committee focussed on GenAI in schools and universities. It also focussed on GenAI tools, especially AI chatbots, that generate text in response to user prompts.
Report outline
1.25This report consists of five chapters, including this introduction.
- Chapter two provides an overview of the use of GenAI in the Australian education system. It considers domestic approaches and international best practice for GenAI in education. It focuses on how to design and integrate Gen-AI education tools in the Australian context; highlighting our policies and guidance, curriculum, and development of product standards. It also examines which rules, guardrails, and regulatory responses may assist in managing its use.
- Chapter three discusses key risks posed by GenAI in education. It looks at safety, wellbeing, and security concerns, such as algorithm bias, overreliance on technology, and privacy matters. The chapter also considers suggested mitigation and protection measures by government, EdTech companies, and educational providers.
- Chapter four examines the opportunities that using GenAI tools can offer in the Australian education system. It considers how the technology could support students and teachers. It discusses the potential benefits of personalised use and support for all students, and GenAI’s additional relevance to marginalised and diverse cohorts. It also discusses the opportunity to bridge the digital divide and improve equity of access to GenAI tools in education.
- Chapter five considers possible effects on the role of educators, the broader workforce and teaching. It highlights the need to build AI literacy and capacity across the education system. Upskilling pre-service and existing teachers is critical to educate students, and their parents and guardians about GenAI. It also looks at issues around assessment, and academic and research integrity, and the need for a strong evidence-base about the effects of GenAI.
Acknowledgements
1.26The Committee would like to thank everyone who provided written submissions, attended public hearings, and hosted and participated in the Committee’s site visits. The Committee extends its gratitude to the expert panel for providing thoughtful and helpful input. The Committee was impressed by the depth of knowledge and enthusiasm displayed by those who contributed to this inquiry.