Researchers at the Massachusetts Institute of Technology (MIT) are currently testing a brand new computer system referred to as Data Science Machine. This new system might one day replace human intuition.

The Data Science Machine is a computer system designed and developed to find patterns in different data sets like a database of weekly profits and promotional sales dates faster than humans.


Since the advent of computers and computer programs, we have seen machine completing some jobs at a greater pace compared to humans. However, still human input is needed for choosing what a computer program must look for in a massive data set, and finding meanings in patterns. This new machine by MIT researchers will be automating this process too.

The Data Science Machine has so far competed with a total of 906 human teams in three competitions and has successfully outperformed as many as 615 of them. While the human teams competing with the MIT system used their predictive algorithms for several months, the new machine by MIT managed to compute predictions in a span of 2 to 12 hours.

For carrying out analyzes, the Machine checks the correlations between different data tables using numerical identifiers. Next, it starts updating these identifiers persistently as it keeps on importing data. These identifiers add up to allow the Machine conduct a series of mathematical operations like sums and averages and try to locate trends in the available data.

The idea of the machine was proposed by MIT student Max Kanter’s thesis. Kanter believes that the Machine might successfully work as “a natural complement to human intelligence.” According to him, it has the potential of expediting the data analysis procedure.

Kanter prepared the thesis with guidance from Kalyan Veeramachaneni and will be presenting it at the IEEE International Conference on Data Science and Advanced Analytics, next week. Veeramachaneni said that this new machine might turn out to be an extremely crucial asset for determining which components in a particular data set must be analyzed for drawing a conclusion.

For instance, MIT only records the performance of its students on online courses and not statistics that will allow them to predict a student’s chances of dropping out. The Machine, however, will be able to deduce a student’s likelihood of dropping out by identifying different variables.