Caitlin Owen
2025: Dr Caitlin Owen of Ōtākou Whakaihu Waka – the University of Otago will develop a more transparent and energy-efficient system for automated machine-learning
Automated machine-learning allows anyone to use artificial intelligence to identify patterns within their data and make predictions about the future. However, current versions of this technology are energy intensive and act as "black boxes," offering no explanation for their predictions. This lack of clarity is a major barrier for applications like health diagnosis or disaster prediction, where understanding the “why” behind decisions is essential.
Dr Caitlin Owen will use their New Zealand Mana Tūāpapa Future Leader Fellowship to pilot a more transparent approach that will allow the user to pinpoint any sources of inaccuracy or error. Instead of "the computer says no", the approach will move towards "the computer says no because...", by allowing errors to be fixed and improvements to be targeted. This approach should require less computational power than current systems, and should also foster greater trust and enable safer adoption of artificial intelligence.
Dr Owen presenting at the Artificial Intelligence Researchers Association Conference. Supplied.