Mr. John O'Connell:

To complement what Dr. O'Sullivan said, the overriding principle has to be human intervention. Whether people are being hired into the sector or their performance is being managed, there has to be human oversight. There has to be a pathway for workers to be able to have a conversation. Everything Dr. O'Sullivan has outlined are the tools and the data, that then feed into decision-making. If there is a bias in the system, it then propagates that on an ongoing basis with no human oversight, or right of appeal. It extends the limitations because it does not have boundaries. For our people who are working in call centres, thousands of whom do so remotely now, the boundaries that existed in a physical space are no longer there. If we have a private conversation, it is a private conversation. However, if I have a private conversation with the Senator and it is being monitored without us being aware, we can find ourselves in a situation of being performance managed, or more serious than that, as a result of that conversation. If people do not know and they are busy, they can have interactions and conversations. From our point of view, it is about human oversight in all forms. No form of AI should have the independence to operate with no human oversight. How that is then used has to be subject to human interaction, appeal and oversight.