Why the Ethics of Algorithmic Decision-Making are so Important
Algorithms are used in various applications across an increasing number of industries. They are used to determine premiums when you take out insurance, and even to decide who will and who will not be given mortgages and loans. Don’t be surprised if your search engine starts offering you some new Hearthstone cards – you’ve searched for something somewhere that has been noted down. Also, when Amazon and other e-commerce stores make recommendations of products we might like, those products have been determined using algorithms. They are used in the online gambling industry, at sites like Discasino, to ensure fair gaming and to maintain return-to-player ratios. We can also see them in online gaming to match players in gaming lobbies. And with the enhanced use of AI, we are coming face to face with algorithmic decisions more often.
But how do algorithmic decision-making and ethics get along?
Algorithmic Decision Making
While we commonly think of algorithms and their use in decision-making as being part of AI, the truth is they have been used for many years, and most of us come face-to-face with them daily. Their use can be quite trivial, as in recommending the next song for you to listen to on Spotify, but algorithms are also used in industries like healthcare and finance, where their decisions can be life-changing.
» Entertainment
When we log on to Netflix, Spotify, YouTube, or virtually any other streaming or entertainment library service, we are confronted with algorithmic decisions. These services use algorithms to analyze your viewing and listening habits, compare them to those of others, and then decide what other content might interest you.
Card collecting will also become different with the increased use of algorithms, as well as other related games. Algorithms that follow personal ratings in Hearthstone Battlegrounds are already old news, and we can only expect more advanced features in the time ahead.
» Finance
Loans and insurance products have long used underwriters to create risk profiles for borrowers. These risk profiles are then used to determine whether an applicant is successful or not and the level of interest or payments they will have to make. Finance companies now use algorithms to assist in the decision-making process or, in some cases, to replace human underwriters. For example, an insurance company will look at a driver’s age and driving experience, and compare this to known accidents to determine how much of a risk they pose.
» Healthcare
Similarly, algorithms are used in healthcare. They can be used to determine ideal treatments that, ideally, improve patient outcomes. Algorithms might also take into account the cost of treatments compared to the likelihood of success.
» Government
Governments, especially departments that are stretched for budget, may use algorithms to help make important decisions. When you apply for a benefit or a grant, it could be a machine that makes the decision of whether you receive, and what level you receive. There have even been some reports of algorithms being used to determine fines and prison sentence lengths.
The Benefits Of Algorithmic Decision-Making
There are benefits to using algorithmic decision-making: benefits that have seen them become so popular:
- Remove Bias – Humans are biased, often unintentionally. Some decision-makers may be more or less inclined to offer loans or mortgages to specific groups of people. Or hospital administrators might accept or refuse treatments based on unknown bias. Algorithms can remove this bias from the decision-making process.
- Expedited Decision Making – Looking through applications on a case-by-case basis takes a long time, and, whether that is in private businesses or government departments, time does cost money. Machine algorithms do not require individuals to work on them at all times and decision-making time can be greatly reduced.
- Reduced Overheads – Because algorithms don’t require people, and the process is sped up, it cuts overheads for the businesses and governmental departments that use them.
- Ethical Decision Making – By removing human bias, and sticking to strict requirements, it is possible to make ethical decisions, although algorithms typically require hygienic data practices and human oversight to achieve this.
The Pitfalls Of Using Algorithms To Make Decisions
There are benefits to using algorithmic decision-making, but there is also the potential for the technology to be misused.
- It Lacks The Human Touch – Algorithmic decisions remove the human touch. That can help eliminate bias from the decision-making process but, sometimes, bias is beneficial. Would you want an algorithm deciding on the best treatment plan for your child, if that algorithm takes into account the cost of treatment? A human doctor or administrator may be less inclined to look at cost as the ultimate factor.
- Determining Accountability – Self-driving cars make algorithmic decisions throughout the driving process. If a self-driving car determines that the best course of action to avoid an accident is to speed up, potentially breaking the speed limit, should the “driver” or the algorithm writer be held responsible? Should anybody be held responsible?
How To Overcome The Ethical Conundrum
There are benefits and pitfalls to algorithmic decision-making, but the fact is that the reduced costs and faster decision-making times of this method mean that it will continue to be implemented in different industries and instances. While an incorrect decision over your next best TV show to watch isn’t that important, ideal treatment plans, mortgage decisions, and grant applications are a lot more important. So, what can be done to overcome the potential ethical problems?
» Human Oversight
Algorithms can be used to do most of the legwork with humans left to make the final decisions. This human oversight can ensure that the right decision, or at least the human decision is made. It also means that accountability exists throughout as part of the process. People are more likely to trust algorithmic decisions if they are overseen by real people.
» Algorithmic Hygiene
Algorithmic hygiene is the detection of bias within algorithms. Algorithms are written by humans, and AI is taught by humans, which means that bias, whether unintentional or not, can also be learned. A hygiene plan identifies this and other potential problems with the algorithmic decision-making process before making recommendations and ensuring as fair a process as possible.
Submit your Top 500 Legend Build, be seen by thousands of people!
Check out the Bazaar "the first free to play Hero-Builder for PC and Mac" Beginner's Guide! https://bazaar-builds.net/the-ultimate-beginners-guide-to-the-bazaar-tips-strategies-and-more/
