Unit rationale, description and aim
Generative Artificial Intelligence (GenAI) is transforming modern computing by enabling machines to synthesise, reason, and act across multiple modalities.
This unit provides an advanced exploration of the models, algorithms, and architectures that underpin state-of-the-art generative and agentic systems. Understanding the characteristics of foundation models—pre-trained large language models that form the backbone of generative AI is essential for identifying their capabilities, limitations, biases, and risks. Students will critically examine how these characteristics influence performance, reliability, and ethical deployment in real-world contexts. The unit covers text, image, audio, and video generation; instruction-tuning; in-context learning; tool use and reasoning with large language models; and the integration of generative components into autonomous and agentic systems. Students will gain applied expertise in the architectures and training dynamics of foundation models, including transformers, diffusion models, variational autoencoders, and reinforcement-learning-based frameworks. Through practical implementation and critical reflection, students will develop the capacity to design, evaluate, and apply generative AI systems that are innovative, transparent, and socially responsible. This unit supports ACU’s mission of ethical innovation and aligns with the UN Sustainable Development Goals (SDG 9 – Industry, Innovation and Infrastructure; SDG 16 – Peace, Justice and Strong Institutions).
The aim of this unit is to equip students with advanced theoretical understanding and practical skills to build and deploy generative and agentic AI systems that are responsible, ethical, and trustworthy.
Campus offering
No unit offerings are currently available for this unit.Learning outcomes
To successfully complete this unit you will be able to demonstrate you have achieved the learning outcomes (LO) detailed in the below table.
Each outcome is informed by a number of graduate capabilities (GC) to ensure your work in this, and every unit, is part of a larger goal of graduating from ACU with the attributes of insight, empathy, imagination and impact.
Explore the graduate capabilities.
Analyse foundational models and architectures unde...
Learning Outcome 01
Apply and fine-tune state-of-the-art generative mo...
Learning Outcome 02
Evaluate the performance, reliability, and limitat...
Learning Outcome 03
Design and conduct experiments to investigate and ...
Learning Outcome 04
Collaboratively document and present generative mo...
Learning Outcome 05
Content
Topics will include:
- Introduction to Generative AI: Foundations and Evolution
- Transformer Architectures and Large Language Models
- Autoencoders and Latent Representation Learning
- Diffusion Models and Text-to-Image Generation: Cultural Appropriation, Visual Sovereignty, and Indigenous Data Governance
- Multimodal and Cross-Domain Generative Systems
- Fine-Tuning, Prompt Engineering, and Evaluation
- Agentic and Tool-Using AI Systems: Collective Agency, Responsibility, and Indigenous Worldviews
- Applied Project Presentations and Technical Reflection
Assessment strategy and rationale
The assessment strategy is designed to develop technical mastery, ethical awareness, and applied innovation capability through authentic, industry-aligned tasks. Assessments progress from foundational implementation and experimentation to applied agentic design and finally to reflective professional synthesis.
The first task builds core competency in implementing and evaluating foundational generative models. The second, a team-based Agentic AI Project, requires students to design and deploy an intelligent agent incorporating retrieval-augmented generation (RAG), orchestration, and responsible innovation practices using contemporary tools using contemporary generative and agentic AI platform The final reflective report consolidates learning through critical evaluation of generative and agentic systems, highlighting accuracy, transparency, and ethical implications.
This sequence ensures students demonstrate independent inquiry, teamwork, accountability, and the ability to connect theory with practice through reproducible, industry-standard workflows, including continuous integration and continuous delivery (CI/CD).
To pass the unit, students must achieve all learning outcomes and an overall grade of 50% or higher.
Overview of assessments
Assessment Task 1: Technical Notebook - Foundatio...
Assessment Task 1: Technical Notebook - Foundations of Generative Models
Students will implement and evaluate foundational generative models to explore latent representations and sequence-to-sequence generation dynamics. As part of this task, students will design and conduct a small-scale experiment (to systematically investigate how these choices influence generative performance for a defined applied task.
30%
Assessment Task 2: Agentic AI Project Stud...
Assessment Task 2: Agentic AI Project
Students will design, develop, and evaluate a generative or agentic AI application that demonstrates model integration, tuning, and evaluation for a defined creative or industry context. As part of their project, students will reflect on how Indigenous knowledges and perspectives on technology, sustainability, and community wellbeing can inform responsible AI design. They will critically evaluate how their solution aligns with principles of inclusion, social responsibility, and respect for Australian Aboriginal and Torres Strait Islander values, contributing to ethical and socially aware innovation.
40%
Assessment Task 3: Reflective Report – Responsibl...
Assessment Task 3: Reflective Report – Responsible Generative & Agentic AI
Students will critically evaluate the accuracy, completeness, and reasoning quality of a generative-AI-produced report on a topic in generative modelling, comparing outputs against scholarly sources and identifying technical inaccuracies or gaps. As part of this evaluation, students will reflect on how Indigenous knowledges and perspectives on knowledge systems, creativity, and responsibility may challenge or enrich mainstream AI-generated reasoning, promoting culturally informed and socially responsible understanding of generative AI.
30%
Learning and teaching strategy and rationale
This unit adopts a research-informed and practice-oriented approach designed to develop both conceptual understanding and applied capability. Learning activities are structured around authentic challenges relevant to deep neural networks, encouraging students to connect theory with practice through independent study, guided exercises, and reflective engagement.
Students will participate in structured learning experiences such as readings, multimedia resources, problem-based tasks that support analysis, critical thinking, and knowledge application.
Representative texts and references
Foster, D. (2023). Generative Deep Learning: Teaching Machines to Paint, Write, Compose, and Play (2nd ed.). O’Reilly Media.
Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.
Ho, J., et al. (2020). Denoising Diffusion Probabilistic Models. NeurIPS.
OpenAI. (2023). GPT-4 Technical Report. arXiv:2303.08774.
Ramesh, A., Dhariwal, P., Nichol, A., & Sutskever, I. (2022). Hierarchical Text-Conditional Image Generation with CLIP Latents. https://arxiv.org/abs/2204.06125
Recommended Readings:
Cropley, D. & Cropley, A. (2021). Creativity and Ethics: A Tension in Generative AI. AI & Society.
Lewis, M. et al. (2020). BART: Denoising Sequence-to-Sequence Pretraining for Natural Language Generation. ACL.
Saharia, C. et al. (2022). Imagen: Photorealistic Text-to-Image Diffusion Models. arXiv:2205.11487.
Vincent, P. et al. (2008). Extracting and Composing Robust Features with Denoising Autoencoders. ICML.
United Nations Educational, Scientific and Cultural Organization. (2023). Ethics of Generative Artificial Intelligence Framework. UNESCO.
Frameworks and Standards:
Hugging Face Transformers, Diffusers, and Datasets.
OpenAI API (GPT, DALL·E), Stable Diffusion, Midjourney.
PyTorch, TensorFlow, and Gradio for interface prototyping.
Responsible AI Toolkit, Model Card Toolkit, AI Fairness 360.