Algorithmic Bias

It’s easy to think machines have some kind of impartiality to them given they are, well, machines. But anything built by humans has a human inside it. Algorithms are no different, and just like us, they are filled with bias.

Algorithm – a word once confined to University mathematics departments and computer labs – now takes pride of place in every second tech news article, determines what you see online and why you received this email at 7am Australian Eastern Standard Time. So it pays to understand what they are, the impact they have and the biases they’re so often driven by.

So let’s go back to the start – what are algorithms in simple terms?

Algorithms 101 – An algorithm is a set of step by step instructions used to do something or make a decision.

With a definition this broad you can see that we humans use them everyday. Even sorting the laundry into darks, colours and whites, and washing them each separately is technically an algorithm. The steps to cook something (a recipe) is an algorithm. Where computers come in is that they can follow a very large number of steps, on a very large data set, and make decisions quickly and precisely. (In the clothing example above – the data set is the clothes and their colours, and the steps are where to sort each piece of clothing).

Algorithmic bias occurs when a computer system reflects the implicit values of the humans who are involved in coding, selecting and collating data to execute the algorithm. The emergent problem with the algorithms in big tech is that they’re designed to achieve corporate outcomes, not societal ones. Their values are simple: to make as much money as possible.

Algorithms now run so deep and cross-reference so much data that what we input has little to do with the outputs we receive. What we now have on the web is ‘the illusion of choice‘. It isn’t just our feeds on social forums which are decided by algorithms. Even what we search for is biased towards corporate algorithmic design parameters. Just search for anything on google you want to buy, in any category and the first bias is plain to see – it assumes you are after the cheapest version possible of every item: clothing, sneakers, airline tickets, hotel rooms, you name it. Apparently we all want to cheapest version of everything no matter what it is. Even if you put the word ‘high quality‘ before the primary search term you’ll still be guided by price. It’s not before we get very specific with words like expensive or search specific brands before we can find what we might need. Another in search is recency bias. Search will always show the latest version, or story of anything and anyone unless a clear time stamp is included.

When we look at social feeds – it’s clear that their algo-game is built on emotional leverage: birthdays, parties, engagements, births, deaths, family events and of course, controversy. These stimulate engagement and keep us on the site longer. Our desire to feel loved, important and often enraged are all that matter to them.

While these examples seem innocuous enough, the proliferation of an algorithm-based society is reinforcing many social biases such as gender, race, ethnicity and economic status to name a few. The canary in the coal mine is dead, the miners are still digging, and yet Silicon Valley are still making bank unfettered. So what should we do?

Like all technology, algorithms are neither good nor bad – they’re just tools. Tools that need to be civilised with some metaphorical workplace health and safety guide rails. They are here to stay, and so our best bet is to make them better. I see two paths forward:

  1. Change – We need to push for transparency on algorithms. Know what’s in them and have to ability to turn them off on demand. I don’t care if the algorithm owners are for-profit corporations – we can and should be able to regulate their output as much as we can a box of cornflakes. No company ever made a decision which reduced its profit until society made it so. It’s time we pushed for big tech to air their dirty laundry.
  2. Opportunity – We need to remember that every flaw in an industry, every broken promise or self-serving design leaves the door ajar for a nimble entrepreneur to make a more respectful version of the product we’ve got little choice on. Algorithm-based tech is no different. Maybe it’s just me, but I think the world is waiting for a social platform we can trust that isn’t designed around extracting unlimited hours and outrage from the people it’s supposed to serve.

Above all, we should never forget that capitalism works best when it is guided by society, not the other way around.

If you like this blog post – forward to a friend who will dig it.