This year I'm looking at how algorithms and the data they process—and by extension, technologies like AI—are used in decision making and how that affects our lives.
One of the first things I'm looking into around this topic is the Australian government's current welfare reform initiative known as the Priority Investment Approach to welfare reform.
Here are some of my notes—I'd love to hear if you have any thoughts. Get in touch.
The initiative comprises two main components.
- An actuarial analysis which attempts to value the future cost to tax payers of all people currently on welfare.
- A set of yet to be determined interventions proposed and run by the non-government sector aimed at reducing that welfare bill (the $96m Try, Test and Learn Fund).
This whole system is modelled of a similar initiative from New Zealand starting around 2011.
I'm collecting reading material and press reports relevant to this topic.
- The focus on cost (of future welfare) potentially incentivises the wrong things.
- Bringing down the future cost of welfare—getting people off welfare—is not necessarily synonymous with an improved life.
- The idea of early intervention and providing more targeted help is at odds with other government policy such as the imposition of a longer waiting period for Newstart.
- The emphasis on cost skews intervention criteria to younger people and those who are easier to help.
- A reasonable proportion of benefits (or cost reduction) may accrue to state, territory or local governments and is not necessarily captured by the actuarial analysis which will be used to evaluate success.
- What are the measures of success and methods used to evaluate programs (see FOI).
- Are individual outcomes tracked for those in groups targeted by programs?
- What are some of the perverse incentives which might result?