Automated decision making in the public sector
A few months ago, I took part in a discussion on law reforms in the context of automated decision making and administrative law. Recently, the Australian Law Reform Commission released their update on the outcomes of these consultations. I recommended we should introduce a language around automation levels in the public sector. Here is my suggestion.
We are already here!
Whether we like it or not, we already are subjects of automated decision making. For some of us, our tax returns are processed without a human casting an eye on them. You might have received a speeding ticket that was generated mostly automatically. Perhaps a police officer reviewed and authorised the decision.
But automated decision making is not equivalent to autonomous decision making. On one end of the automation spectrum, we might have algorithms that provide us with information relevant to our decisions. Close to the other end, humans are merely authorising and triggering a decision that was fully prepared by a computer system.
The automotive industry is dealing with the different levels of automation all the time. There, six levels of automation are commonly used: Level 0 to Level 5. They are easy to remember: no automation, hands-on, hands-off, eyes off, mind off, and full automation.
We should introduce an equivalent classification in administrative law. What if we could talk about different levels of automation of administrative decisions? Imagine seeing a label on every decision you receive that tells you how automated the decision process was: no automation, assistance, recommendation, human-approved decision, human-monitored decision, and autonomous decision.
What do you think about it? Could a consistent language around decision automation help us avoid some of the challenges we have seen in the past? Would it provide more clarity? Could it help us reimagine the process of disputing a decision?
Hit reply to let me know what you think!
I am growing this newsletter. If you think it might be interesting to someone in your network, please let them know. Thank you!