Last week, I attended and presented at AI Day, New Zealand’s premier artificial intelligence event in Auckland. The variety of speakers from different industries from finance, to housing and education, was a clear indication of how AI really is everywhere.

We heard from companies who have developed facial recognition and digital assistants, and learnt about the impact conversational AI will have on day to day life. AI assistants are already being developed and used in a healthcare context, from voice-enabled virtual assistants, to chatbots that can converse with patients. At the moment, they serve as an invaluable tool for helping clinicians make the most informed, precise and timely decisions that they possibly can. 

David Heiner, Strategic Policy Advisor at Microsoft spoke about the intersection of AI and ethics, presenting a framework of 6 pillars to consider when using AI technology. These were applicable to any industry using AI, and aligned closely to what we’re starting to see implemented in the health data science space in New Zealand. He summarised by saying we should be building AI technologies centred on human values, to ensure safe and reliable outcomes.

As health is such a risk-averse industry, it was particularly interesting to hear Anchali Anandanayagam discuss the legal perspective on AI. For the health industry, the future could quite possibly see AI used to make decisions, so it is important that we start asking questions such as who is ultimately responsible when machines are making these decisions? My perspective is that there should always be a human involved in these decisions, but it is still important that we plan for when things might not go as the human planned.

Sean Gourley, founder and CEO of Primer discussed some of the current ways that algorithms are being used, which was fascinating, but also a little unsettling. He raised the question, as we teach machines to think for themselves, are we prepared for what they are going to reveal? Or more pertinent to health, how do we make sure the machines ask the right questions. One example might be in a risk prediction model for a health issue, the algorithm could focus on the wrong defining factor, throwing the predictions off completely. In this scenario, unless the human trained the machine with full transparency in mind so they can understand why the machine made the predictions that it did, it could compromise patient safety.

The theme of ethics was prevalent throughout the conference, which shows that it’s front of mind when building AI technologies. It was reassuring to see all these different companies and industries thinking about it and more importantly working through how we can improve on the development and regulation of these technologies so that we are prepared for the future with AI.

To learn more about how algorithms might be used in the future of health, read my blog below.