Jenny's take on the important issues of letting a computer decide
Concern with unconscious bias is not a new thing, it’s been around a long while. I recall many of my guardians suggesting I put myself in the shoes of others to consider other perspectives.
A recent SheSaysMcr meetup on unconscious bias has stirred me into action to share my data views on our retail environment.
Even if you were raised to consider the viewpoint of others, it’s difficult as an adult to challenge our entrenched values and systems. This is because our first computer – the brain – has two operating systems. The fast part of the brain creates impulsive thoughts. Christina Connolly advocates that you are not responsible for your first thought. You are responsible for your second thought and first action. It was remarkably reinvigorating to hear from someone, at the forefront of technology in retail space, talk about how conscious they are not to project their preferences into their design.
As humans we are a global society battling with our relationship with technology, cultural heritage and emotional well-being. We are hectic with the internet of things and our dependence on technology makes us cyborgs by definition.
The devices and apps we use day in and day out are learning from our data to shape our future. So it’s critical for us all to understand how human activity is interpreted by data, and how the computer learns. If we are to participate in a revolution – that celebrates diversity, nurtures variety and reverses the trend on mental health and the increasing polarisation of wealth, health and equity – we need to be data savvy.
I am a big believer in sustainable meaningful data that facilitates better decision making. I think it’s paramount that digital leaders help businesses and education to design systems that gently nudge you into considering deeper truths.
Let me tell you about some immeasurable human influences that have been affecting my retail data.
I have been using data in retail to influence supply and demand for 12 years, which in digital terms is a very long time. I recognise the limitations of the data and know it’s only going to be indicative. I also recognise that humanity and market forces are at play. The computer science does help me to qualify trends and lead me to insightful conversations.
In 2007 I had to ring a Tesco store in Birmingham to find out why our best kidswear range was not selling well in their store, despite the store being a top performing store for clothing sales. I will always remember the clothing manager’s response. “Oh, they love the range,” Pauline said, “but they know it gets marked-down in price after several weeks of launching so they wait for the price drop. If they can’t wait till then, they pinch it!”
I was reminded of this story at Christmas 2018 whilst supplying some high-ticket items to Asda. We identified a new top-selling store and allocated them more stock and chose this store to visit to see the but upon a store visit found that the shelf was not well-stocked. The store assistant said they were not putting the stock out, to keep the shrinkage numbers to a minimum. The shop assistants were minimising theft at the expense of sales. They were insensitive to the fact that the global replenishment system, that allocates stock, will repeat that trend and so inadvertently program the sales to decline.
Last week I went through a till at a big supermarket and witnessed a teenager on the checkout giving away clothing items to a woman with a child. I asked if he had been given discretion to do that. Part of his response was to explain how his dad had struggled so much as young father, that now when he sees families that look like they are struggling, this modern-day Robin Hood gives them the occasional unscanned items. Well how do you measure that?
So my point here is, there are things at play that we are not measuring and so it's hard to influence them. These factors are eating into the algorithms that will help play out the future. It’s critical therefore that we design systems that help us see what is at play and consider the human impact of tech advances.
It’ frustrates me to see so many sales assistants have been replaced by self-checkouts. Then stores have had to employ security guards to stand around whilst customers decline a receipt and walk out of the shop whilst the unmoved security label is beeping and no one bats an eyelid.