It seems like algorithms increasingly rule the world. Only this summer, the debacle around the A level results demonstrated what far-reaching consequences the use of algorithms can have. How did algorithms become what they are? Might algorithms soon replace human decision-making altogether?
First, the word ‘algorithm’ simply means a structured scheme to solve a problem. For instance, there is a structured scheme to solve a puzzle: take one piece at a time, and add to another piece. Some puzzles can even be solved blindly in this way. Similarly, the algorithm to make Spätzle — my favourite German noodle dish — is: mix 250g flour, 2 eggs, 125ml water, and some salt; then, push through a coarse sieve into boiling water. I am sure you can think of many algorithms from your everyday life.
Early on, people tried to formalise algorithms. Some of the earliest formalised algorithms were found on clay tablets from Babylonian mathematicians, dating from about 2500 BC. The clay tablets were found near Baghdad, which was also home to the influential Persian polymath Muḥammad ibn Mūsā al-Khwārizmī. Looking closely at his name, you’ll notice how the Latinisation of his name and some creativity yield the word ‘algorithm’.
Over time, the study of algorithms became an academic discipline in its own right: computer science. Whilst it bears ‘computer’ in its name, what this subject truly studies are systematic ways to solve problems: algorithms. Rather than thinking about solving problems, many researchers in this field think about what classes of problems can or cannot be solved. You may have heard of the P vs NP problem. This problem is of such importance that those who solve it will be awarded 1 million US dollars.
Algorithms are hardly a new concept. They have always been with us. By themselves, they are nothing more than a structured set of operations. An algorithm is transformed into action only when executed by a human being. Indeed, the ability to craft tools — to derive systematic ways to solve problems — is deeply ingrained into human DNA. This is nothing more than coming up with algorithms. One could say: I algorithmise, therefore I am.
Many recent developments have shown how algorithms can affect society and individuals. Whilst it may seem that algorithms are to blame, it will always be humans who design and deploy algorithms, and who either fail to see the consequences of their adaption or willingly accept the potential harms of such tools. Instead, we must seek to avoid the potentially harmful consequences of algorithms, just as we must do with all our decisions.
The good news about algorithms is that they can be changed easily, as opposed to humans. Algorithms could help us overcome our inherent biases and weaknesses, by supporting human decision-making, rather than attempting to replace it. Indeed, any such attempt is doomed to fail, because algorithms will always be rooted in decisions taken by humans.
There are millions of different ways to design algorithms for a given problem. Let’s make sure we use the right algorithms.