It is obvious why anyone would want to put a cork on corruption and then throw it into space to disintegrate itself. So, when a group of scientists from the University of Valladolid in Spain put together a computer model that can predict instances of graft in government agencies, the whole world took notice.
Here are some of the top takeaways from the study that was published in FECYT - Spanish Foundation for Science and Technology. It was first published online in Springer on 22 November 2017.
What was the study about?
According to the research paper published in Springer, the study created a computer model based on neural networks that would send out warnings for possible instances of graft occurring in a government office. These warnings can then be used for corrective and preventive measures, which in other words, means changing the way a government functions or weeding out certain bad apples, for lack of a better term.
The model in the study uses corruption data extracted from several provinces in Spain where graft occurred between 2000 and 2012. A lot of different factors were at play that defined how a routine incident of corruption would occur in any given government agency.
The researchers’ aim was to understand what factors play a key role and then administering changes to those factors in an attempt to eradicate corruption. Of course, this process will be iterative, as no “bad habit” can be weeded out in a single go.
What were the findings?
According to the paper published by the researchers, the following are the key takeaways. Since Spain follows a customized form of parliamentary monarchy, it can be easily translated and adapted for other similar governments. Of course, the data will need to be updated.
- Public corruption is a cause of multiple factors such as:
- Taxation of real estate and a steady increase in property prices
- Nature of economic growth, GDP rate, and inflation
- Increase in the total number of non-financial institutions
- Sustenance of any single political party for a “long time”
- Since data on actual cases of corruption was used to create the model, it provides for a better look at the factors compared to how the model would have been had it depended solely on the perception of corruption. Such studies have not yielded much either have they garnered any interest from the public
- Corruption can be predicted three years before they are bound to happen
Where does AI come in the picture?
Since all of this sounds too good to be true, it is wise to ask what the role of artificial intelligence is in this study. In order to do that, let us go back to other studies that have tried to predict corruption.
All of the previous studies on the subject have depended on data that were more or less subjective indexes of perception of corruption, reports Science Daily. What this means is that the only type of data being used is something that is available on the public domain. Since the government can sometimes come in between the sourcing of this data by private agencies like Transparency International, the database stops becoming useful. All it will give the model is data that does not reflect the true gravity of the situation.
On the other hand, when actual data is used, there is much for artificial intelligence to feed itself and then bring out a model that can be used to predict the very nature of corruption. This is where this new study excels when compared with historical reports.
The biggest role of artificial intelligence in this exercise is to find a correlation in a set of data through a process that attempts to mimic the human brain functionality. If we feed human being scores of detailed court cases about corruption charges on government employees, he will take decades to dig through them and still end up without an actionable conclusion.
A neural network, on the other hand, analyses the data, studies its various factors, and creates a relationship between them to see what connects with what. Let’s take a rough example to make this clear.
If out of 10 cases of corruption, 8 of them involves a particular modus operandi and a similar cause for the exchange of money to take place, then such a connectionist system will flag that as a recurring factor. This information, along with several others, is then used to reach a conclusion. Of course, this example is an imaginary one, and the amount of data fed in the study were a lot higher, which makes the result even more constructive. Over 12 years of data of actual corruption cases are bound to give such an AI tool enough ground datum to work with. But, it should still be noted that no amount of data is sufficient when one is trying to execute the predictive analysis.
Finally, according to the Chr. Michelsen Institute, this type of predictive tool that depends on patterns can well be a smart anti-corruption tool. Its ability to handle big data and detect anomalies in it is what makes it a promising new system that could be adopted by late the 2020s. However, it also points out the single biggest concern over its use: it will force for more surveillance in the world as data about even the smallest corruption cases will be added into the system. Data that will include personal details of individuals.
It is a great relief that AI in theory at least does not mimic the dangers portrayed by science-fiction films and is being seen as a technology that can help humanity lead a better life. Its usage in anti-corruption neutral networks is a step in the right direction, and with added research, it will pave way for better governance. Where those being governed have one less accusation to make against the government.