Miss Kiowa Scott-Hurley1, Dr. Chris Hines2
1DSTG, Australia, 2Monash eResearch Centre, Australia
Biography:
Kiowa Scott-Hurley is a Dja Dja Wurrung woman who spends her 9-5 leading a team of awesome Machine Learning engineers at the Defence Science and Technology Group. She's currently studying a Masters in Cybersecurity, and her unique mix of skills across ML, Cybersec and HPC means you can always catch her grumbling something about GPU drivers. In her free time you'll find her covertly drawing comics during boring meetings, using LLMs for evil, and getting nerd-sniped by malware analysis.
Abstract:
In the whisker-twisting world of artificial intelligence (AI), it is vital that responsible and ethical practices prevail. However, purr-suading stakeholders to embrace Responsible AI frameworks can feel like herding cats – and with such a breadth of frameworks available, it can feel impossible to get started right meow. We propose using cattitude to shed light on the importance of transparency, explainability, accountability, fairness, and privacy in AI development.
Art poses a memorable and accessible way to communicate complex and serious ideas to a breadth of audiences. In this talk, we offer a cat-comic based overview of the NIST (National Institute of Standards and Technology) framework for implementing responsible, explainable, and ethical AI. We’ll also discuss the hiss-conceptions and challenges of implementing these frameworks in the eResearch environment, such as the abstract nature of most frameworks, and the difficulty of realistically meeting these guidelines when funding timelines are short.
Whether you're a curious kitten or a seasoned alley cat, this session is a unique crash course in preventing cat-astrophe when developing AI systems.