When you hear the word “algorithm” what comes to mind? You might think of the facial recognition technology in our phones, or self-driving cars. Or you may think of the mysterious technology responsible for providing an endless stream of cat videos on YouTube.
But what exactly is an algorithm? And how can they help support a more equitable Aotearoa?
For those of us out of the know, the word “algorithm” can be somewhat intimidating. We might see them as black boxes, comprehensible only to math whizzes and those with computer science degrees. However, algorithms are a lot simpler than you may think.
An algorithm is a process or set of rules that can be followed to solve a problem or make a decision.
There are many things that could be considered an algorithm. A recipe for baking cheese scones, a step-by-step manual for assembling a table, or the directions provided to you by Google Maps that outline the quickest way to get to a destination are all examples of algorithms.
Simply put, algorithms are a set of clearly defined instructions that can be followed to complete a task. This principle holds true whether you are a human executing an algorithm in the form of a recipe, or a computer executing an algorithm in the form of code.
With the rapid advancement of computer technology over the last few decades, algorithms have played a major role in transforming the way we live, work, do business and process information. Their power to handle high volumes of data at incredible speeds has improved the speed and efficiency at which we operate across almost every facet of society.
Algorithms also provide a unique opportunity to create more equitable systems and services for people across Aotearoa.
Government departments in Aotearoa have begun to lean more on algorithms to support decision-making, craft policy and deliver services (Nicholson Consulting have helped build many of these algorithms!). Algorithms can improve the speed at which government services are delivered to people, make it easier to identify people in need of assistance and help make decisions in a fair and consistent manner.
A great example that illustrates this is the algorithm we built for ACC, which helps fast track the assessment of the 2 million claims they receive every year. Prior to this every claim would be assessed by an individual person. This meant that it could take up to 2 weeks for patients to know if their treatment would be covered. Now patients can lodge their claim and the algorithm will assess it based on data such as the diagnosis, age and keywords. If the claim is straightforward it will be approved instantly and if it is more complex it will be referred to a human. With the help of the algorithm, patients are able to know the outcome of their claim within a matter of seconds. This not only places patients in a better position to start their recovery process, but it allows for ACC to focus their resources on the people who need the most help.
You can read more about our work with ACC here
He Ara Poutama is another great example of algorithms in action. He Ara Poutama is a tool that we developed that can help government agencies and policy makers understand how to reach 1 million te reo Māori speakers in Aotearoa by 2040. With the help of an algorithm, He Ara Poutama uses data to forecast how different interventions will impact the number of te reo speakers in the future. This information will be used to inform investment decisions and contribute towards the revitalization of te reo Māori.
You can read more about our work on He Ara Poutama here
As well as building algorithms, we’ve also done mahi that relates to algorithms at a governance level. Working with the Ministry for Social Development (MSD), we developed the Model Development Lifecycle, that details the approach to assure decision makers at MSD that the appropriate technical, legal, ethical, and te ao Māori risks and opportunities have been well managed during algorithm development. The Model Development Lifecycle is freely available and aligns to the Stats NZ Algorithm charter. You can take a look at it here.
While algorithms are a powerful tool that can help government agencies and other organisations work towards more equitable outcomes, they are not without risks.
Algorithms are only as good the people and systems that create them, and as a result can end up embodying the biases of those that produce them. This is especially important to keep in mind for algorithms that operate on a large scale and/or impact the lives of vulnerable or marginalised communities. Failing to consider the impacts for all groups and populations can result in inequities being embedded into systems that people engage with and rely upon.
The problem of bias is exacerbated when you consider the issue of transparency. The process through which an algorithm reaches a certain decision is not always clear to your average person. While this can be an issue for even simple algorithms, technology such as machine learning can make the decision-making process even more opaque. This is especially an issue if an algorithm produces an outcome that seems unjust. This lack of transparency can make it difficult to challenge unjust outcomes as the reasoning behind them is not clear.
In order to use algorithms in an equitable way, it is important to mitigate their risks while also harnessing their strengths. At Nicholson Consulting, we recommend thinking about the following things when undertaking a project involving algorithms:
1. Understand how humans and algorithms can work together
Algorithms are great at performing simple tasks at speed. This means that they can process these tasks faster and at higher volumes than humans alone. Humans, on the other hand, are great at complex decision making and can navigate scenarios that are novel, or not so straightforward.
The best outcomes are reached when humans and algorithms work together.
Algorithms should be treated as a tool to help people make decisions and not as a replacement for human decision making. By leaving the simple tasks to algorithms, humans can direct their attention to where it will be most beneficial. This allows for systems and services to meet people’s needs more efficiently, while also ensuring that complex decisions have been handled with the care they deserve.
Part of using algorithms in an equitable way is maintaining transparency around how they operate. This is especially important for algorithms that have a significant impact on people’s lives. Transparency is essential for maintaining trust and accountability in the systems that use them.
Part of making algorithms transparent is making them explainable. The process through which an algorithm arrives at an outcome should be understandable to all those who build, use and are impacted by them. This ensures we know whether the algorithm is functioning as intended and outcomes that seem biased or unfair can be properly challenged.
One important way of doing this is by retaining human oversight. Keeping humans in the loop provides an avenue for redress, as there will be someone who can be held to account for outcomes produced by algorithmically driven systems.
3. Centring people and their wellbeing
Finally, building equitable algorithms involves consideration for the wellbeing of all those who will be impacted by it. From design and development through to final implementation – people’s wellbeing should be prioritized at every stage.
To do this it is important to take a systems perspective. This involves fully investigating how an algorithm may create inequities for different populations and working towards mitigating these. Part of this includes maintaining transparency, accountability and human oversight as mentioned above; however, this can include other things such as:
Consulting with affected groups
Keeping track of whether the algorithm is functioning as intended
Considering any privacy or ethical issues resulting from the way data is sourced or used
Algorithms have transformed many aspects of the way we live our lives. They also provide an incredible opportunity to create systems and services that are more efficient, transparent and equitable. Through understanding how to balance the risks posed by algorithms with the unique potential they offer we can work towards creating a more equitable Aotearoa.
To learn more about how we use algorithms to build a more equitable Aotearoa reach out to us at email@example.com