The Future of Elections: Can Artificial Intelligence Predict Poll Results?

Polling in the United States has come to an end, with Trump beating Hillary Clinton to the astonishment of the "audience" around the world. All of the pollsters, who with their sophisticated surveys and mathematical models, finally went disastrously wrong. At the same time though, someone got it right. And, it came from none other than MogIA, an artificial intelligence (AI) program built by an Indian startup, Genic.ai. Its analysis was based on reviewing public engagement on 20 million data points from renowned platforms like YouTube, Google and Twitter.

MogIA, named after Rudyard Kipling's fictional character in "The Jungle Book," just didn't guess this election rightly. It did so with both the Republican and Democratic primaries in 2016 and with the last three US Presidential elections.

However, these results haven't been publicly verified. Creator of the AI, Sanjiv Rai, had earlier told CNBC that if Trump loses, it will disregard the data trend for the very first time ever since internet engagement started in full swing.

While a majority of algorithms suffer from biases from the end of the programmer/developer, MogIA strives at learning things from her immediate environment and develops her own rules, said Rai. She doesn't discard any data, which is why she has turned out to be an expert, added Rai.

The artificial intelligence system was built in 2004 in Mumbai, and has been getting smarter with time. Using social media to predict the results of elections has become extremely popular because of the huge amount of information that can be easily accessed.

Nick Beauchamp, a professor at Northeastern University, had experimented with artificial intelligence to over 100 million tweets in the election of 2012. The results, according to him, closely reflected the actual outcome. In an era, where self-driving vehicles will be hitting the roads soon, it's likely that a major chunk of our everyday affairs will be managed by smart AI systems like MogIA.