MACHINE LEARNING FOR DESIGNING FOR FINANCIAL SECURITY, SPRING 2018
B. LAREAU, C. PRINZ, L. SCHERGER, A. STAHL, K. THOMAS, K. WHITE

INTERVIEW DATA:
“My parents didn't teach me much about savings.” 

“I learned small things from him [my dad] but not really more than I should've.”

“I learned saving on my own, my mom just said save money and don't spend it.” 


MISSION:
Enable parents to prepare their children for financial independence using allowance as an educational tool

WHY/HOW/WHAT:
Smart Allowance dynamically introduces rewards and incentives to teach children how to handle their finances responsibly; an educational experience as unique as themselves
Purchased spending data

USAA member spending/saving data



Cluster analysis of spending/saving behavior combined with domain expert advice

A/B test rewards/incentives for kids



DATA:
  • Purchase a data set about how parents allocate money for their children
  • The user’s previous spending habits - what the child has spent and what the parent has given the child
  • Pre-training will need to be done on the advice giving system to identify categories of spenders and give personalized
  • Removes the human work of creating and curating sample budgets. Only one to two screens to adjust settings for the parents. Simple ML curated choices are presented

ETHICAL QUESTIONS:
  • Are we fostering a dependence on an application they might not have in the future?
  • What if the app gives poor advice to the parents or gives the wrong advice for a certain situation?
  • What if the app never reaches an ‘equilibrium state’ of healthy spending and kids are getting cycled through different spending profiles?
  • Who decides what good advice is?
  • What happens if the software provides really bad advice?
  • Is there a way to game this system to cheat out rewards with suboptimal spending habits (ex. Spending poorly one week and very well the next to claim a reward that requires huge changes in progress)
  • Are we breeding children that will become obsessed with money/greedy?
  • Does our product focus so much on rewards that no attention is spent on actually learning