You may have seen a
In the process of preparing that article for publication, I searched for a stock image to run with the story. After searching for “layoffs,” I tried “pink slip.” I got several options that fit the bill – dismayed men and women clearing off their desks, angry bosses holding literal pink slips and the like. And then there was this:
Technically, I suppose, this is accurate. And, to be clear, it was but one result amid a few dozen, which also included photos of pink slips – the undergarment sometimes worn beneath a dress or skirt – and one shot of a child in a pink outfit on a playground slide.
What struck me most, however, was uncertainty as to whether this was the result of a clever algorithm or a wrench in the works of machine learning.
Week in and week out we hear credit unions, CUSOs, tech vendors and more talk about the value financial institutions can find in AI and machine learning. These technologies can provide numerous benefits to CUs, but the people touting the value of those systems frequently follow up with a second component that’s just as important: a human touch.
AI and machine learning can help credit unions fight fraud, for example, but it still takes an MSR to reach out directly to the member to double check on a suspicious transaction or unusual purchasing patterns. Machines aren’t infallible, and it will take dedicated credit union staffers working alongside this technology to fight fraud.
Because, as my stock photo experience illustrates, sometimes the machines get it wrong – even if they’re technically right.
