Madgwick Student

Academic Integrity is guided by the Student Academic Integrity Policy. A breach of Academic Integrity includes presenting under the student’s own name work substantially written by someone else. “Someone else” includes artificial intelligence tools.

The roll out of Madgwick for Students does not change UNE’s policies on use of AI in assessment. Where use of AI is permitted, Madgwick for Students provides a safer and more secure platform than others you might use. Whether use of AI is permitted will vary depending on the course and unit you are enrolled in. If you have any questions about the use of AI for your course or unit, you can speak to your unit coordinator.

Students are not obliged to use AI except where it is a requirement of a particular unit or course. This tool is provided to them if they choose to engage with it. 

In certain assessment contexts, UNE may require the use of Madgwick Student. However, more generally, UNE recommends that where use of AI is permitted, students choose to use Madgwick for Students because of its security and safety benefits.

Students use existing UNE login credentials, once they have been invited to join. They will receive an email when it’s their turn to enrol, and this will be over a three-month period from November 1 2025.

In the same way that students and staff are responsible for verifying information obtained from a Google search, they must also take responsibility for the outputs generated by AI systems like Madgwick. The phenomenon of AI hallucination—where an AI generates incorrect or misleading information—is well-documented and cannot be entirely eliminated, despite ongoing advancements. 

Understanding AI Hallucination

1. Definition: AI hallucination refers to instances when an AI model generates outputs that are factually incorrect or nonsensical, often presenting them with high confidence. This can occur due to limitations in the training data or the model's inability to understand context fully. 

2. Prevalence: Research indicates that AI hallucination is a common issue across various AI models, including those developed by leading organisations like OpenAI and Google. 

Best practices to mitigate risks

To address the potential for incorrect information, we implement several best practices: 

1. Quality Models: We provide access to high-quality AI models from reputable organisations such as Google, Anthropic, and OpenAI. These companies conduct rigorous testing for hallucination and regularly publish their findings, ensuring we use the most reliable models available. 

2. Training and Education: We offer training for students and staff focused on best practices for using AI tools. This includes grounding AI outputs in credible sources, developing strong research habits that emphasise verification, and encouraging critical thinking when interpreting AI-generated information. 

3. Verification Protocols: We advocate for a culture of verification, where users are expected to cross-check AI outputs against trusted sources before utilising them in academic or professional contexts. This mirrors the responsibility expected when conducting traditional research. 

Madgwick is a custom tool for UNE, built by Simtheory, that provides access to AI tools in a secure environment. Unlike the commercial AI platforms, Madgwick does not use user information to train AI LLMs. The platform is hosted exclusively on cloud infrastructure within New South Wales. All model inference and data processing remains within NSW boundaries.

Yes, Madgwick meets accessibility standards (https://www.accessibility.org.au/guides/what-is-the-wcag-standard). 

Madgwick is a platform that provides access to a range of general purpose AI models to support a range of tasks. 

Cogniti is a specialised AI platform designed specifically for educators. It provides a secure and intuitive environment where educators can harness AI tools to enhance their teaching practice, create engaging learning materials, and provide timely feedback to students. Unlike general purpose AI tools, Cogniti is built with education in mind, ensuring equitable access to advanced AI models while grounding responses in course materials. 

Madgwick is a closed environment hosted exclusively on cloud infrastructure within New South Wales. All model inference and data processing remains within NSW boundaries. There is full encryption of both data at rest and in transit across all platforms. 

Your chat interactions in Madgwick Student are private. They are not used to train AI models and are not visible to your lecturers or tutors. However, some limited information may be accessed by UNE’s technical support team if required for troubleshooting, safety monitoring, or to review content that has been flagged by the system’s safeguards. Any review is handled in line with UNE’s privacy and security protocols.

You should only upload material that you have the right to use. This includes:

• Your own work – draft essays, notes, reports, or other original content you’ve created. 

• UNE-owned documents – lecture notes and resources, transcripts and any other teaching related content relevant to your unit that you have legitimate access to.

• Public domain works – material where copyright has expired, such as classic literature.

• Openly licensed content – works under Creative Commons or similar licences that explicitly permit copying and adaptation.

• Small, cited excerpts for a specific purpose – pasting a short section (e.g. a few paragraphs) from an article to summarise or critique, consistent with “fair dealing” for study and research. (This does not allow uploading entire documents or books.)

For further information see UNE Copyright policy

Simtheory is a Newcastle-based tech company focused on building novel AI solutions. Simtheory is also the name of the company’s key platform, which provides an interface through which customers can work with most of the key AI models and use them interchangeably. The platform also supports the creation of AI assistants. 

UNE initially approached Simtheory because of its innovative approach to AI for higher education.  

“Madgwick” is UNE’s branding of its own custom Simtheory platform, named for UNE’s first Vice-Chancellor, Sir Robert Madgwick. 

UNE has established a dedicated team to drive AI adoption, called LabNext70. It is driving innovation by developing AI-driven solutions that enhance the learning experience for our students, assist academics, and empower staff. 

LabNext70’s mission is to develop novel AI technologies in a meaningful and ethical way that could be described as ‘humanised technology’, because its objective is to give people more time to do meaningful things.  

Yes. Different AI models require different levels of computing power, which means their environmental impacts vary. UNE students will be provided access to Gemini 2.5 Flash and Claude Haiku 4.5, models that are designed to be lighter, faster, and more efficient for everyday tasks like summarisation, planning, or drafting text. By using a right-sized model for common study needs, students can complete their work effectively while supporting UNE’s commitment to responsible and sustainable AI use.

UNE is committed to reducing our energy use and on the Armidale campus our own solar farm (opened in 2020) supplies an average of 38 per cent of the total campus electricity (academic and Bellevue campuses combined). During summer that can increase to 50 per cent of the total campus daily consumption.