How can flags help us understand (and govern) AI?

27 August 2021

What do flags and artificial intelligence (AI) have in common? Patterns. When you take a closer look at flags, it’s easy to notice certain colours, shapes, and symbols that create patterns. AI also detects patterns in nature, languages, images, human interactions, and many other fields. 

In this article, flags will help us demystify AI, understand its logical basis – pattern recognition – which is as old as human attempts to understand nature and life in general. AI amplifies this ‘simple’ technique of pattern recognition through big data and enormous processing power of modern computers.

By clearing the air about what AI is and isn’t, we can have informed and balanced discussions on AI governance and the future of this much hyped technology.

Pattern recognition is behind search engines, speech recognition used by smart devices, fingerprint recognition, x-ray analysis, and many other areas of AI application in modern life. 

Before we dive deeper into patterns and AI, let’s talk about the (sometimes overlooked yet monumental) importance of flags for humanity

Flags mobilise us for sports competitions, protests, wars, and even lynchings. People kill each other under and for their flags.

Apart from their use in geopolitics and wars, flags are part of our daily reality.  You can find them on buildings, shopping malls, and tourist venues. In international centres like Geneva and New York, where the representation of countries is intensive, flags are an integral part of the everyday landscape of the city.  

Flags are one of the most visible symbols of our belonging to social groups. Through flags, we represent ourselves as, for example, fans of a sports team, members of a local club, and nationals of a country. 

Even looking at the past few months, we can see the colourful pieces of fabric which hold such massive meaning for people in the centre of a variety of events.  At the Tokyo Olympic games, the world’s best athletes competed behind their national flags.

On an entirely different context, flags were in the spotlight in August as well, as the Taliban entered Afghanistan’s capital Kabul and flags were one of the first and most visible symbols of their control of the city. 

​​

Patterns in flags and AI 

 

Pattern recognition which is used a lot in machine learning turns messy dots into patterns. 

It is exactly what we will do in the next few examples when a melange of colours and shapes of colours turns into a categorisation of flags.

Most flags have common patterns shaped by history, culture, and identity as explained in the excellent infographics by the Danish company Ferdio, featured in this article. 

The main grouping of flags is around colours. The most dominant colour in national flags is red, while purple isn’t used in any flag. Red mainly symbolises blood. Blue represents the sky, the sea or rain. Green symbolises vegetation and fertility. It is also used in the flags of Islamic countries, as the colour of Islam which dates back to the Fatimid Caliphate. 

 

The most colourful flag is South Africa’s national rainbow flag, which unifies the Dutch settlers Boer’s red-white-blue flag with the yellow-red-green of the African National Congress.

 

On the flip side, there are many countries with only two colours in their flags.

For the full range of colours from two to five you can visit the infographics. Different colour patterns emerge in different continents and regions. 

In Oceania, the dominant colour is blue, which is not surprising considering that the region is mostly composed of many small island states surrounded by oceans. 

Most African flags are inspired by the flag of Ethiopia, which dates back to the 1890s. The Ethiopian flag of green, yellow, and red was also a source of inspiration for the Rastafarian movement which started in the West Indies and spread worldwide. 

Flag families display belonging to a certain culture or nationality. For example, Scandinavian flags include the symbol of the cross, dating back to the Danish design of the Christian cross. 

Slav nations combine blue, red, and white – a legacy of Peter the Great, Russian emperor (1682-1725). The Russian flag was a by-product of the establishment of the Russian Navy by Peter the Great. He was impressed by the Dutch navy, which generated the inspiration to use the blue, white, and red of the Dutch national flag. Many other Slavic nations have been using the same combination of colours. 

For Arab countries, the combination of red, white, green, and black colours emerged during the Arab revolt in 1916.

Flags are just one example of patterns. Patterns are all around us. They are in nature, achirecture, arts, language, and our daily routines. 

Flags, patterns, and AI governance

 

The baseline principle guiding AI is not very different from categorising flags: it is a matter of recognising patterns. For example, under the bonnet of powerful Google searches are AI algorithms that identify patterns in the ways information is organised or what and how people search for things. Facial recognition is all about patterns as well. While other complex problems need more data and processing power, the underlying logic remains similar whether we’re dealing with mapping flags or the human brain.

As patterns are identified and used in algorithms, a new and more problematic question emerges: the ownership of patterns. Recent developments point in a scary direction: not only do tech giants control such powerful technology but they also have ownership over patterns codified in algorithms.

As patterns and ways of discovering them via AI are privatised, we may hamper the future of human creativity. Patterns should remain a public good open for human creativity and scientific inquiry for generations to come. We should pass this common heritage of humankind to them as previous generations made it available to us.

The first step in ensuring that AI serves as an enabler of human creativity is to try and understand what AI is about as we tried to explain via this analogy between flags, patterns, and AI. 

See more:

Subscribe to Diplo's Blog