As Donald Trump became the 45th President of the United States of America defeating Hillary Clinton, his campaign not only defied expectations and conventions at every turn, but also proved all predictions wrong. All but ONE!
The 2016 US presidential election results may have come as a shock to Democrats and a great surprise to many Republicans, sending after shock waves throughout the world. But for Sanjiv Rai, a first-generation Indian innovator and serial entrepreneur, it came as a confirmation of the abilities of his AI software MogIA. While all TV analysts and media experts in the US had mistakenly given Hilary Clinton a 5.2 lead, MogIA accurately predicted that Donald Trump would still be in the lead and win by a roomy margin.
The election results on November 8, 2016, took all of America by surprise as Trump romped home with 306 electoral votes to Clinton’s 232 votes. The counts were later adjusted to 304 and 227 respectively, after defections on both sides, formalizing Trump’s election to the Presidency.
What is most interesting is MoglA predicted Trump’s victory in October, even before the FBI announced it was examining new Clinton emails following WikiLeaks revelations about impropriety.
MogIA is an artificial intelligence system and election predictor, and has now successfully predicted its fourth election in a row. The artificial intelligence system, created in 2004, has been learning from its environment and getting smarter each day. The system, which learns in real-time by analyzing data on the internet, placed its bets on Trump in October to win Presidential Election after it successfully predicted both Democratic and Republican party primaries.
The term ‘MogIA’ has been derived from the name ‘Mowgli’, the child from Rudyard Kipling’s novel The Jungle Book. Rai thought this was the most suitable for his AI model because it learns from the environment.
Where did this confidence come from?
“The AI system shows that the candidate in each election who had leading engagement data ended up winning the election. If Trump loses, it will defy the data trend for the first time in the last 12 years since Internet engagement began in full earnest,” Rai had explained in an interview to CNBC in October. If someone was searching for a YouTube video on how to vote, then looked for a video on how to vote for Trump, this could give the AI a good idea of the voter’s intention.
To be fair, Rai admits there are limitations and MoglA can’t always analyze whether a post is positive or negative. Nonetheless, it has been right in predicting that the candidate with the most engagement online wins. “If you look at the primaries, there were immense amounts of negative conversations that happen with regard to Trump. However, when these conversations started picking up pace, in the final days, it meant a huge game opening for Trump and he won the primaries with a good margin,” Rai explained.
How MogIA scored over others?
It would be interesting to note that when it came to popular vote, Clinton beat Trump by 2.1%. Her nearly 2.9 million votes advantage is the largest raw total among candidates who did not win the Presidency. This is the fifth time in US history, and the second time this century, a Presidential candidate has won the White House while losing the popular vote.
But Clinton was lagging far behind Trump when it came to Electoral College votes. This is because the popular vote does not determine the winner. To win the Presidential election, a candidate must receive a majority of Electoral College votes. This means a candidate must have votes distributed all across the country instead of winning a few heavily populated areas, like Clinton did with California. As Pew Research pointed out, take out California, and Trump wins even the popular vote.
For instance, in this election the state of Wyoming cast about 210,000 votes, and thus each elector represented 70,000 votes, while in California approximately 9,700,000 votes were cast for 54 votes, thus representing 179,000 votes per electorate. Obviously this creates an unfair advantage to voters in the small states whose votes actually count more than those people living in medium and large states. But that is another debate. What matters is how did MogIA get it right where all other heavyweight American analysts and experts failed?
Where Rai scored over other analysts is the way MogIA sifted through location data of millions of social media users. While most other surveys and polls focused on the total numbers to predict a Clinton win, MogIA was smart enough to know that it was not the total numbers that mattered but how the votes were spread across the country to deliver states to each presidential candidate. Predictably, while analyzing the social media shares, MogIA took into account each post’s locations, and it artificial intelligence capability could compute how a certain number of votes from one particular region could make the difference, instead of the same number of votes coming from another area.
In the end this is what mattered — while Clinton was leading the popular vote by a sizeable number, it was Trump’s popularity, spread evenly across the country, and not just the populous states, which led to the shocking victory.
Artificial intelligence has advantages over more traditional data analysis programs. Most algorithms can get influenced by a programmers or developer’s bias, MoglA on the other hand develops its own rules at the policy layer, and develops expert systems without discarding any data, says Rai. He believes that the system can be improved by more granular data, for example, if Google gave MoglA access to the unique internet addresses assigned to each digital device. Considering there’s huge amount of data available online, using social media for predictive analysis is likely to become popular and has a long way to go.