Although IBM WATSON PREDICTS NARENDA MODI WILL WIN THE HONOR, THE UNU SWARM DISAGREES
On Wednesday, TIME Magazine will name its 2016 Person of the Year. In advance of the highly anticipated announcement, TIME partnered with Opentopic and IBM Watson to predict the winner of the award. As TIME explains, “Opentopic has evaluated more than 62 million documents from nearly 3.5 million online sources. Watson’s deep-learning technology then classifies and sorts that information” to determine the most likely candidates, and a massive online Readers’ Poll was then conducted to name the public’s predicted winner.
At the end of this complicated process, the large poll picked Indian Prime Minister Narendra Modi its likely winner in a relative landslide. Modi received nearly 18% of the vote, while his nearest competitors, Barack Obama, Donald Trump and Julian Assange received no more than 7% each. It’s important to keep in mind that TIME’s Person of the Year is awarded to the person who “affected the news most, for good or for ill, and embodied what was important about the year.” TIME famously named Hitler the Person of the Year in 1938, while Stalin won in 1939 and 1942.
With that criteria in mind, is Narendra Modi really the most likely recipient of TIME’s Person of the Year? To find out, researchers at Unanimous AI asked UNU, a Swarm Intelligence platform that allows groups of online users to think together as a “brain of brains” that can express itself as a singular unified entity. The group that made the Person of the Year prediction was comprised of 75 randomly selected individuals from around the world, connected online by real-time Swarm Intelligence algorithms. Swarms have proved to be eerily accurate in the past, outperforming the experts at the Oscars, the World Series, the Stanley Cup, and the Kentucky Derby, where the Swarm correctly predicted the Superfecta, turning $20 in nearly $11,000.
The Swarm was asked to narrow down TIME’s list in an iterative process that identified the least likely person to win. Unlike the Opentopic and IBM Watson predictions, once the list had been winnowed down to the final six contestants, the Swarm felt that Modi was actually the least likely person to win TIME’s award.
Looking at the graph below, you can see how the Swarm made up its mind – initially torn between all the choices, but quickly converging on the conclusion that Modi is in fact the least likely of the six top finalists to win the big prize. Known as a “faction analysis,” such graphs help to show how a swarm is not a poll — it’s system, that deliberates and converges upon optimal answers.
In contrast the CRISPR scientists, Flint Whistleblowers and the two major party nominees for President of the United States are all deemed much better candidates for TIME’s award than Modi. Soon the list would be pared to three contenders: Clinton, Trump, and British Independent Party leader Nigel Farage, who championed the BREXIT movement.
In the end, Clinton and Trump dominated the Swarm’s discussion the way they have dominated public conversation for the past year. And, so it was only fitting that they would face off one final time. And, once again, Donald Trump got the better of Clinton, being named by the Swarm as the most likely person to win TIME’s Person of the Year. As you may remember, Trump was runner-up for the award a year ago, and his new status as the President-elect makes clear that his impact will only grow from here.
As you can see, the Swarm reached its prediction in a matter of seconds. So, unlike the IBM Watson prediction which involved extensive polling and crawling millions of websites and documents and predicted Narendra Modi as the winner, UNU’s unique Swarm AI needed only a relatively small group of 75 people and a matter of minutes in order to register Donald Trump as the person who “affected the news most, for good or for ill” in 2016.
So, which team will get it right – Time/Watson or UNU? All we know right now is that the Time/Watson approach, which involved polling a massive crowd, predicted Modi, while the UNU intelligent swarm predicted Trump. In prior comparisons of “crowds vs swarms” the swarm has come out ahead, so our bet is on Trump. You can read a formal academic study of “crowds vs swarms” here.
We’ll find out on Wednesday which approach was more accurate.
Want to make sure you don’t miss any of the Swarm’s insights? Enter your email below and we’ll keep you up to date with the latest news, views, and predictions!