Today we take a quick look at the bias of artificial intelligence and the data gap that contributes to that bias.

Artificial Intelligence is far from neutral

The concept that AI is biased is not new to me. Sometimes it’s hard not to notice how biased it is. For example, do a Google image search for “grandma” and the majority of images will be of white women. Google dictation accurately delivers the spelling of Dormammu (a fictional Marvel character), but can’t get perimenopause right. AI is not neutral because white men are generally the ones who design it. They use data collected by white men about white men. More on that below.

However, what I learned from Coded Bias is the designers who create the algorithms don’t understand how the algorithms work! They feed data in, something happens, and then data comes out. It’s often unclear which data sets are used and if they include women and people of color. Leaving a huge gap that can dramatically impact people’s lives as AI becomes more prevalent in more settings.

I highly recommend watching Coded Bias (it’s on Netflix). It features a number of brilliant women and their work to combat the impact of biased AI as it creeps into our lives.

Things I regularly use were not designed for me

More fascinating tidbits from Invisible Women that validate my experiences and help me feel like I’m not crazy. There are two things that happen to me on a nearly daily basis:

1. I can’t open some kind of jar or packaging. Then, I spend the next 10 minutes wondering how the fuck I will open something when I get older if I can’t open it now, and how the hell older people get their jars open.
2. I bemoan that my right hand/elbow/thumb hurts from doing things on my phone.

Thankfully, the book explained why that is.

Criado-Perez states “Women also have on average a 41% lower grip strength than men, and this is not a sex difference that changes with age: the typical seventy-year-old man has a stronger handgrip than the average twenty-five-year-old woman.”

She also says that women, on average, have smaller hands than men. That means women, on average, also have a smaller hand span. A difference that would be significant to the design of things like PPE, tools used for agriculture, and our phones! This has an impact. For studies on smartphones that sex-disaggregated their data, there is a “statistically significant gender difference in the impact of phone size on women’s hand and arm health.”

It is not unlikely that my phone was not made with me or my hand size in mind. It’s also not surprising that I would be feeling the impact on my arm and hand. Though she doesn’t say it in the book, it also makes me think that cans and jars (and the tools we use to open cans and jars) were also likely not made with me in mind.

Get that jar open

A running theme in Invisible Women is that even when data shows that women may be impacted adversely, the response is rarely to change the design. Instead, the response is to change the woman. Meaning that I have no hope that somebody will create jars that I actually can easily open any time soon. Instead, I offer these tips:

1. Run any jar with a metal lid under hot water. This makes the metal expand and can make the jar easier to open.
2. Break the seal. Get yourself an old-fashioned can punch bottle opener and use it to lift the jar lid open just enough to break the air-tight seal. You can usually see and hear the pop when this happens. Once you’ve done that, it’ll be much easier to open.