Media reports say an artificial intelligence (AI) system called MogIA, developed by Sanjiv Rai, an innovator based in Mumbai, has predicted that Donald Trump will win Tuesday's presidential elections in the United States. Unveiled in 2004, the system apparently got it right in the last three presidential elections. It also predicted that Trump and Hillary Clinton will be the nominees of the Republican and Democratic Parties respectively.
Rai is quoted as saying that the algorithm got even better as it has “learnt” from the last few rounds. MogIA is named after Mowgli from The Jungle Book. Like the child in Rudyard Kipling's endearing tale, MogIA learns from its environment. MogIA's judgement is based on the analysis of millions of interactions on Facebook, Twitter, Google and YouTube relating to the two candidates. Skeptics question whether information from social media can accurately predict popular political preferences in elections. In any case, new technologies like artificial intelligence, on which MogIA is based, are already being used to influence social media interactions.
The campaigns of both Trump and Clinton, for example, have chat bots that dump massive numbers of tweets in very short periods, creating the illusion of a “trending” opinion. According to researchers from Oxford University, nearly one-third of tweets favoring Trump and one-fifth promoting Clinton between the first and second election debates came from handles run by robots — together, they produced more than a million tweets.
It does not really matter whether MogiIA's pick, Trump, comes through in what has become a very tight contest in the last couple of weeks. Rai's algorithm certainly picked up the late momentum in favor of the Republican nominee. Clinton's expansive lead of nearly 12 points, in some polls, after the third debate on October 19 appears to have melted quite a bit in the last few days.
What is important, however, is the fact that the technologies underlying MogIA — the ability to process vast amounts of information in quick time and the capacity of computers to learn and improve their performance from practice — are transformative and here to stay.
This revolution involving AI has already begun to change the way election campaigns are run in the US. Computing power has long been used to appreciate trends in public opinion and the specific nature of voter concerns across different locations and demographics. These judgments helped campaigns to better deploy time, energy and resources in reaching targeted audiences with the right message.
It is well-known that President Obama's campaigns during 2008 and 2012 had put these techniques to good use. Rapid advances in processing big data and machine intelligence have now made it easy for all campaigns — massive cross-country presidential ones to local elections — to take advantage of new technologies.
Not everyone is happy with the political use of AI. There is a growing concern that AI is amplifying the negative dimensions of modern political processes and sharpening polarization by spreading falsehoods, demonizing opponents, promoting tensions between different social groups and stoking xenophobia.
All this is true. But it was also the case with earlier communication technologies, like the printing press, newspapers, radio and television. All of them expanded the power of mass communication and have been put to both positive and negative use. The scale and speed of AI is indeed something else.
That AI is very much part of our political life is underlined by the plans of American mainstream media to use bots to cover the results of the campaign on Tuesday night. Robot journalism has been around for a while. Computers have been writing stories on quarterly earnings of corporates and covering sports events. This week, the use of AI in American newsrooms will scale a new peak.
If news reports are, in essence, about turning raw data into a narrative text, bots are getting good at it. CNN, NBC, The Washington Post and The New York Times are all deploying bots to improve the speed and efficiency of their coverage of results from hundreds of elections, besides the presidential race, across America and delivering them to consumers on smart phones. Use of the bots, the editors hope, will free up the journalists from number-crunching and let them focus on interpretation.
The influence of AI on the electoral processes may be significant; but it pales in comparison to its deep impact on the economic bases of modern politics. If the so-called fourth Industrial Revolution, built around AI, eliminates jobs, both blue and white collar, at a rapid pace, some of the negative political trends we have seen in the US elections could acquire a sharper edge. The consequences will not be limited to developed societies. If AI delivers new means of production, it is bound to alter relations of production within nations and between them.
As advanced societies debate AI, Delhi seems strangely passive. Prime Minister Narendra Modi deployed some of the new technologies in the 2014 campaign and has recognised the importance of the new technologies for India's development. Yet, Delhi is finding it hard to translate the PM's instinct for the digital into effective policies on AI.