The more data, the more effective AI will be

Many people see AI as a futuristic technology that has little impact on their daily lives. Nothing could be further from the truth! Today’s deep learning technologies, for example, stem from work done by Frank Rosenblatt as early as 1954!

Only two years later, a machine designed by Arthur Samuel was already proving capable of beating a grandmaster at checkers. It took longer for machines to become more successful at more complex games like chess and go, but they eventually dethroned humans in these disciplines too, in 1998 and 2016 respectively.

Symbolic AI, statistical AI

Generally speaking, an AI is a computer program that analyses data and makes decisions of an “intelligent” nature on its own. There are two main classes of AI: symbolic and statistical. Symbolic AIs solve problems on the basis of a set of rules encoded by a human being. This can be a machine that plays chess, chatbots, video game engines, etc. or even a temperature, pressure or speed regulator. Overall, they work on an “if… then” basis: “if” a player enters a certain area of the game, “then” a certain event will occur. Research on this type of AI has not changed much since the 1970s.

At present, statistical AIs are the focus of attention. They are able to make decisions on the basis of a statistical study of data. It is the data itself that gives rise to the operating rules of these AIs. The more data there is, the more effective the AI will be. This data can be of any kind: images, data from a company’s customers, measurement values in scientific research, etc. Autonomous cars such as the Google Car, for example, use opaque statistical AI to drive them.

Very concrete applications

AI is now used on a global scale in a large number of business sectors: retail, telecoms, banking, industry, etc. For example, our company’s AI tool has been able to detect and prevent post-natal complications, reduce the number of infants needing intensive care by 20% and detect, with 95% accuracy, those at risk of heart problems within 6 months. In the future, the deployment of user-friendly self-service tools like TIMi will gradually eliminate the need to dedicate human staff to repetitive and time-consuming tasks such as manipulating Excel files.

What about TIMi?

TIMi is the undisputed world leader in the field of Automatic Machine Learning. Developed since 2007, it allows AI to be created without heavy infrastructure and in just a few minutes. With a complexity comparable to Excel, it is a self-service tool that is not reserved for IT specialists, but can be used by any professional within a company.

One of the difficulties often encountered by our clients is handling the large volumes of data that allow these tools to reach their full potential. These volumes are measured in databases of millions or even billions of rows. This large volume can involve significant costs. This is why TIMi integrates a data manipulation tool (ETL), free of charge and today recognised as the best on the market in the context of Big Data problems.