An artificial intelligence (AI) system that has rightfully predicted the last three United States Presidential elections is now predicting that Donald Trump will eventually win the elections on November 8 and will ultimately take the oath of office on the 201th of January, 2017. Well if you’re a democrat, don’t frown at this, it’s the system that was developed by Sanjiv Rai who is the founder of Indian start-up Genic.ai that says so.
Created in 2004, MogIA collects user engagement data from Google, Twitter, Facebook and YouTube and uses this to predict who’ likely to win the race to the White House eventually. Another interesting thing that MogIA tell us is that Donald Trump’s engagement numbers has now overtaken the Barack Obama 200 numbers by 25 percent.
But this seem to contradict a lot of traditional and online polls that have come out recently which shows Hillary Clinton ahead. That said, even recent polls are showing Donald Trump eating well into Clinton’s lead and if you’re one of those rooting for Donald Trump, this may just cheer you up as the election day draws near.
Now the thing to know here is this, user engagement is a broad one and it all depends on how you choose to define it. So the ordinary data this system uses is mentions. How accurate are mentions. A mentions could well signal support or opposition depending on context. But speaking to CNBC, Rai said Just because somebody engages with a Trump tweet, it doesn’t mean that they support him. Also there are currently more people on social media than there were in the three previous presidential elections.
On the definition of user engagements, he continued by saying “If you look at the primaries, in the primaries, there were immense amount of negative conversations that happen with regards 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”. Going by this statement, it means even negative engagement could still mean a win for the republican who has been in the news lately for negative reasons and has continued to insist that if he loses, it would mean that somebody somewhere may have rigged the election; an allegation even republicans have come out to condemn.
But the big thing to learn from here is that even negative perception can be turned into positivity. Trump has been in the news over the past year for mostly negative stuff but that didn’t stop him from winning the primaries. It also means that with more reporting on Donald Trump, he has been able to advantage of the media (social or otherwise) coverage – good or bad to further an agenda.
On why this system seems to contradict other polls, Rai says he thinks it’s a wakeup call to Clinton to be complacent and that if she ends up winning the race, “it will defy the data trend for the first time in the last 12 years since Internet engagement began in full earnest”.
Artificial Intelligence systems rely on data and the more data available over time to them, the better they get which is why this system has been able to make better predictions over the course of the last three elections in the U.S. On why Donald Trump’s user engagement data seemed to be well above Barack Obama’s, it’s because there are way more people online today than 8 years ago.
Limitations
MogIA is not able to learn about the sentiment behind posts and like I said earlier, it’s all around mentions and mentions may not be enough to accurately predict a race that’s as close as this. In traditional polling, the caller is able to judge more accurately how a voter feels about a candidate and that result can be collected and used to form an opinion.
Opportunity
Having been successful at this for the past three elections, this system is capable of correctly predicting this one as well. While it may limited with respect to human sentiments, it boasts of a larger pool of data to sample. It is able to learn from its environment correctly to get better and this means, the more accurate systems like this become, the more traditional polling methods phase out.
