The algorithm will be able to predict the risk of possible escape of defendants based on their criminal records and court records. This system would help judges decide whether a defendant should wait at home or in prison.
As we announced last year from the Blogthinbig.com, one of the technological trends that will be protagonist in this 2017 will be the growth of the artificial intelligence with its algorithms in most aspects in which it is developed. This includes machine learning, a branch of technology whose objective is to create systems that learn, that is, to achieve algorithms capable of predicting future behaviors as accurately as possible.
A new application of this technology puts an end to a long queue in judicial processes. In particular, it would help judges decide whether an accused should wait at home or in prison until the trial is held, depending on the prediction of the particular case. The algorithm developed by the U.S. National Bureau of Economic Research (NBER) will be able to predict the risk of possible escape of defendants based on their criminal records and court records. More than 100,000 cases within the New York city were analyzed, and their predictions turned out to be more accurate than the decisions of the judges themselves. This fact is a success in the application of machine learning within the judicial world, according to one of the researchers, the professor of computer science at Cornell University, John Kleinberg. However, there are some gaps or” biases ” with the data that are extracted for their predictions. First because they are based on records of sentences handed down by previous judges. And secondly, the variables that judges handle are infinitely greater than those of the algorithm. For example, judges may be concerned about racial inequalities or specific crimes rather than the general risk of crime.
The behaviour of the algorithm focuses on classifying defendants under a risk score based on data from judicial documents, the specific crime, place and date of arrest, as well as the age of the defendant. Apart from the parameters analyzed, for example, the race of the detainee is found. In fact, ProPública last year, a similar system that tried to predict which convicts deserved parole, unfairly predicted high risk for people of color versus those who were not. Even so, the group of computer scientists and economists in charge of the project have calculated that they would be reduced a 25% crimes committed by defendants at large awaiting trial; and a 40% on the prison population awaiting trial. A system that neither intends to replace the work of judges nor prevent their work, since it would only send messages alerting about those decisions that it considers potentially erroneous. In parallel, an analysis of the performance of judges is produced, establishing a prediction of those who most often release accused persons who are more likely to escape or commit a crime again.
Predicting crimes through Big Data
Let us remember that this is not the first national security initiative. From telefónica and together with Nuria Oliver, scientific director of Telefónica I + D in Barcelona, the Crime Hot Spots project was created. It was concluded that Big Data is effective in predicting human behavior, whether it involves a risk to society or not. To do this, they mainly used data collected through smartphones that save the experience of users who have been victims of a criminal act. A data analysis that has enabled the creation of an algorithm that allows future crime scenes to be detected with an accuracy of 70%.
But how are these algorithms created?
An algorithm is a routine that follows instructions. That is, a list of commands that follows established steps. The computer universe is the most faithful in following these types of rules, and algorithms, therefore, become its best tool. Algorithms solve problems highly demanded by companies such as knowing the personalized recommendation of content, predicting the volume of sales, the date and choice of that certain amount of purchases, in order to get as close as possible to the perfect producer for each consumer. The increase in this need on the part of the business sector has the effect that our society has become the civilization of the algorithm. According to El País, algorithms have arrived to take over the economy. In fact, the philosophy of this technique is fully established in our day to day. We each have a personal library with our favorite and recommended content, our perfect catalog.
The social and cultural power of machine learning
Innovation in machine learning it has increasing social benefits. E-Commerce uses this system to thoroughly study their customers and offer increasingly personalized products. One of them is digitalmeteo, which studies meteorology as a phenomenon that impacts customers and their consumption routine, whether specific products or leisure in general. From the BlogThinkBig.com we tell you that this application generates a social environment adapted to you and your needs, for example, imagine that today you decide to make a bike route, but it starts to rain. So that you do not stay idle, you will be offered, for example, alternatives according to the time in your area.
On the other hand, in the last 50 years, science has not been able to deal with a disease that is the world’s leading cause of death: a total of 800,000 people commit suicide per year according to the WHO. But Artificial Intelligence puts a before and after to this reality, creating an algorithm capable of predicting with 80% accuracy the possibility of a person committing suicide during the next two years.
But not only are algorithms born with benefits applicable to the health field. Culture also benefits from them. In fact, and from the BlogThinkBig.com we are already talking about the birth of what is known as expressive artificial intelligence. These “cyberartists” understand the art that we have created and respond to it with new self-creations. The supercomputer IBM Watson with its algorithm called Beat he has created the song “NotEasy” developed by the system under the records of the most successful songs in history.
Meanwhile, other algorithms like BestSellerCode are able to predict whether a novel will be a bestseller or not. This reduces the risk of uncertainty that artists suffer at the moment of launching their work.
If you lie they know
Not only do ICTs participate in practically all areas of our daily lives, but they dare to challenge us and uncover our deceptions. No actress or actor is free of this algorithm that detects more accurately if a person lies or tells the truth, being a very useful tool during court proceedings. This application proved to be even more accurate than the human beings themselves who conduct the interrogations. The creators of the algorithm recorded their data through the platform TheInnocent Project, created in 1992 to help prisoners who were declared not innocent by relying on DNA analysis. For example, some of the witnesses analyzed through the algorithm revealed that their 88 confessions were false.
What algorithms remain to be discovered? What revelations will they bring with them? A Machine Learning and its applications offer so many possibilities that they only depend on our creative capacity. Without a doubt, we live consuming information formed by data, and under this philosophy eCommerce innovate by leaps and bounds. Welcome to the century of algorithms.