Artificial intelligence March 2016

This month we take a look at the world of Artificial Intelligence and its growing presence in daily life.

Google’s AlphaGo wins big

As technology pervades every aspect of our daily life, the role of artificial intelligence is becoming the hot topic for gadget-evangelists and technophobes alike. ‘Computer-learning’technology has been around for decades now, but it’s only now that we’re really beginning to understand the potential impact it could have on society.

In a startling example of how far AI has developed, a Google artificial intelligence programme won a resounding victory over the renowned Go master Lee Se-Dol. Famed for its complexity, the 3,000-year-old board game requires intense concentration and the ability to see several moves ahead at all times. But Google’s AlphaGo programme bested Mr Lee by 4-1, suggesting that algorithms and complex software are developing at a rate that could render many intricate human skills obsolete within a few years. AlphaGo points not only to how far AI has progressed in the past decade but to how quickly it can improve its own understanding. A game that has been honed by the human brain for almost 3,000 years was mastered by AlphaGo over the course of just a few years, with no end to this mechanical evolution in sight.

The success of AlphaGo represents the most significant breakthrough in computer programmed learning since 1997 when IBM’s Deep Blue supercomputer conquered the world chess champion, Garry Kasparov. Demis Hassabis, Chief executive of the company behind AlphaGo, Google DeepMind, was quoted as saying “one day [these algorithms] can be used in all sorts of problems, from health care to science.”
Artificial intelligence March 2016
Go master Lee Se-Dol looks on as AlphaGo takes the upper hand

‘Tay’ sets a dark tone for the future of AI

A less encouraging milestone achieved by AI, Microsoft launched its AI chatbot earlier this month. ‘Tay’, as the programme was affectionately known, was designed to learn from the humans it interacted with through Tweets and DM’s. Microsoft launched ‘Tay Tweets’ claiming the account became more intelligent the more it was used: “The more you chat with Tay the smarter she gets”.

The software did indeed show a remarkable ability to mimic language structure and adapt to human influence. Sadly, Tay’s digital education wasn’t necessarily a cause for celebration, largely because the teachers she interacted with chose to feed her some extremely questionable material. Within 24 hours of going live, the AI software had turned from a rather rigid, albeit friendly, automaton, into a ‘Hitler-loving sex slave’ with a penchant for Donald Trump quotes and deeply disturbing 9/11 conspiracies. The chat robot, which was modelled to speak like a teen girl, had to be deleted after her replies became increasingly un-PC.
 Artificial intelligence March 2016
Tay, the easily led genocidal robot

Incredibly, Tay isn’t Microsoft’s first attempt at a teen-girl chatbot. Xiaoice was intended as a friendly female companion to provide advice to people using Chinese social networks. With around 20m users globally, Xiaoice has been met with significantly less criticism than her younger digital sibling, although that may be because she has yet to resort to alarming displays of xenophobia and troubling sexual innuendos.

Leave a Reply