Dr Penny Stewart

Chief Executive Officer, Petra Data Science

Dr Penny Stewart is a mining engineer, Fellow of the Australian Academy of Technical Sciences and Engineering and founder and CEO of mining AI software company Petra Data Science. Penny worked on mine sites for five years in Tasmania and Western Australia before obtaining her Queensland 1st Class Mine Managers’ Certificate of Competency. She commenced PhD studies at UQ’s Julius Kruttschnitt Mineral Research Centre, where she received the Ian Morley prize for Best Minerals Engineering Postgraduate student. In 2009, she instigated the application of self-organising maps (neural networks) for the study of SLC recovery analysis.  In June 2015, Penny founded PETRA to extract value from mining data.  PETRA is a world-leading provider of machine learning and optimisation solutions to the global mining industry, including their flagship MAXTATM digital twin value chain optimisation suite. In 2016, PETRA collaborated with Newcrest Mining to develop and deploy some of the mining industry's first machine learning algorithms. 

Dr Penny Stewart's keynote title and synopsis:

Holding AI technology to account

AI in our personal lives shapes our perceptions of AI and what it can and can’t do well, along with our perceptions of the “trustworthiness” of artificial intelligence-based technologies. The term “trustworthy AI” has been defined by various governments around the world as a set of principles for AI that are almost impossible to measure against and are aimed to engender human attributes to what is essentially code written and deployed by humans to achieve a set of objectives. Arguably using a distinctly human characteristic such as “trustworthiness” in the context of AI has come about due to the conflation of human intelligence with being human. In our personal lives we have seen social media and other tech companies obfuscate lines of accountability for the impacts of AI algorithms on humanity arguably by personifying AI. This has led to distrust of AI that many of us bring to the workplace that isn’t associated with other types of mining technology.

In contrast, ten years’ experience developing and deploying hundreds of machine learning models into mining operations globally has demonstrated that requirements for mining AI technology are actually the same as for other mining technology. Specifically, the mining industry requires accurate, reliable and maintainable technology that has been properly tested and developed, whilst also meeting relevant engineering standards and legal requirements. In practice this doesn’t mean mining companies need to blindly “trust” AI, it means they need to trust the people involved in creating, testing, and deploying the AI technology, as well as having the necessary commercial agreements in place to ensure the safety, security and legality of AI technology.