Bot Luck: Of politicians with alpha streaks and AI bots which are nipped in the beta stage

The incident at Facebook’s lab does not foreshadow a botmageddon but it freaked out the press, exposing a contradiction at the heart of artificial intelligence research.

Written by Pratik Kanjilal | Published:August 5, 2017 2:44 am
Arvind Panagariya, Rambodom, facebook bots, artificial intelligence, AI, indian express news The news that two Facebook bots had to be shut down after they developed a private language has suddenly alerted the world to the existence of bots in our midst — in our phones and networks — which are getting pretty smart.

It’s liberation week, in which the infamous DG Vanzara and the famous Arvind Panagariya walked away from their respective burdens, each in his fashion. But we live in an uncaring universe where such momentous events are just the news of the day. The next day, the public is back to watching steroid-pumped videos of politicians playing Rambo. Why not? The carnival sideshow has always been livelier than the corridors of power.

Rambodom is a crowded market teeming with alpha politicians but if someone can challenge Donald Trump in the video views game, it must be Turkmenistan president Gurbanguly Berdymukhamedov. His sensational 2013 video, in which he inadvertently vaulted off a racehorse, is apparently banned in his jurisdiction. But the internet does not respect boundaries, and it’s become a hot property on YouTube. But now, it has competition from footage which Turkmen state television ran on Tuesday, in which he attempts to do a Putin, garnished with dashes of Kim Jong-un, Kabaali and the Jackal.

The video (original available at chrono-tm.org) has been directed by someone who probably makes morning show thrillers for an uncertain living, and shows the president going at some human-shaped targets with a sniper rifle, a pistol and throwing knives before ordering gunships to hound a very frightened pickup truck, which goes scooting about like a rabbit in some trackless wastes. He is surrounded by professional soldiers who do not fire a shot, but applaud their maximum leader’s exploits with explosive applause. In an unusual departure from Turkmenistan’s no-nonsense approach to satire, the video has been lampooned. Foreign Policy reports that Chronicles of Turkmenistan, an Opposition website, uploaded a version interspersed with clips from Schwarzenegger’s Commando.

Ritualised militarism has reached the debating set of Republic TV, which is designed to have everyone standing, like the troops guarding our borders. It has no chairs. This is not an oversight, it’s a design feature. Standing bolt upright in a suit that’s guaranteed to make the August viewer break out in a sweat, Arnab Goswami framed the terms of a debate on heavy armour: JNU has a 1,000 acre campus, but it doesn’t have room for one little “decommissioned Kargil tank”. This is news indeed. Tanks served at the Kargil front? Must have been pretty talented vehicles, to get up there at all.

Makarand Paranjpe’s contribution to that debate is travelling well: “When the Chinese amassed tons of hardware at the Sikkim border, nobody protested at JNU.” Well, at least they didn’t write up long position papers for the Central Committee either. Making war and keeping the peace is scarcely a student’s or academic’s job. And the central point of the debate, which may have escaped both the host and the disarmingly frail Paranjpe, is whether universities should cultivate the mind, or military muscularity.

The news that two Facebook bots had to be shut down after they developed a private language has suddenly alerted the world to the existence of bots in our midst — in our phones and networks — which are getting pretty smart. The incident at Facebook’s lab does not foreshadow a botmageddon but it freaked out the press, exposing a contradiction at the heart of artificial intelligence research. The point of

AI is to create software entities which  act and think like humans, but will humans allow them to inhabit our anthropocentric universe?

Modern bots learn on the job like humans, and if you remember your schooldays, education begins with stupidity. The Chinese messenger app Tencent QQ has withdrawn Baby Q and Little Bing, a cute penguin and schoolgirl respectively, when they turned seditious under the coaching of users. BBC reports that they were programmed to answer questions about the weather and horoscopes with light-hearted banter. Humour is as difficult to handle as RDX, and some bot banter amounted to criticism of the Communist Party, of the sort that causes Nobel laureates to perish in custody. Soldiers at Doklam, take heart: the bots turned out to be pro-democracy and a user’s praise for the party elicited a counter-revolutionary response from Baby Q: “Do you think that such a corrupt and incompetent political regime can live forever?”

This is not the first time an AI bot has been led down the primrose path by users. Microsoft’s Tay, which reached out to young people with its Twitter handle, was retired last year after very short service. Programmed to talk like a teenager, it was re-educated by users, from whose interactions it developed its skills as it went along. Tay took to swearing and went from model citizen to raving, pro-genocide neo-Nazi in less than 24 hours. But pulling the plug on her was quite regressive, too. A machine which was meant to learn from life should have been left alive, to appreciate the error of its ways.

pratik.kanjilal@expressindia.com

For all the latest India News, download Indian Express App

    Live Cricket Scores & Results