Artificial intelligence: Cause for celebration or fear?

Browse By

Since its conception in the 1950s, the phrase “artificial intelligence” has come to take on many different meanings. Some find the idea of computers thinking for themselves an incredible scientific feat, while to others this notion conjures up dystopian visions of evil robots intent on destroying humanity.

Indeed, Elon Musk, the billionaire entrepreneur famous for dreaming up projects including SpaceX and Hyperloop, described artificial intelligence as “the biggest risk we face as a civilisation”.

While the future of this technology is shrouded with uncertainty, what is clear is that it is developing at an inconceivable rate.

On 6 December 2017, the world witnessed one of artificial intelligence’s greatest achievements yet. AlphaZero, a new program developed by Google’s DeepMind division, played 100 games of chess against Stockfish 8, generally regarded as one of the top engines on the planet.

To gain some context, know this: Stockfish is programmed with thousands of previous games, and can, therefore, call upon millennia of human experience of the game. What’s more, its processing power allows the engine to examine up to 70 million positions per second.

Meanwhile, AlphaZero only has the capability to consider 80,000 positions per second and was given nothing but the rules of the game. No strategy, no database of past games. In fact, AlphaZero was given just four hours to play against itself prior to the contest.

The outcome of the match was shocking. AlphaZero won 28 games and drew 72, losing none. This result understandably came as a massive surprise to many, prompting Grandmaster Pete Heine Nielsen to say, “I always wondered how it would be if a superior species landed on earth and showed us how they played chess. Now I know.”

But how did a rookie produce such a dominant victory over a proven champion?

DeepMind uses a technique called “deep learning”, which involves continual self-improvement by learning from mistakes, in a similar way to humans. In this way, AlphaZero was able to quickly figure out what works and what doesn’t work, many times over, until it was able to pick out the best move pretty much every time.

AlphaZero was not the first program developed by this group – in 2015, AlphaGo beat a human professional for the first time ever and, more recently, Google’s artificial intelligence has bested some of the top arcade gamers in the world.

So what can this technology do for us? The number of applications is limitless. Artificial intelligence can be used in all sorts of fields, such as speech recognition, self-driving cars and even stock market forecasting.

Of course, there will always be a fear of machines outsmarting us and turning against us. But a lot of this is all down to speculation, and the first thing scientists will consider is maintaining control over their programs. With correct regulation and monitoring, how could we ever give computers the chance to put us in any real danger?

It seems as though, given enough time, there is nothing that artificial intelligence cannot achieve. Surely then, we should recognise this technology as an absolute necessity for technological development? I would certainly say so. The only problem we humans will have to handle is making sure the computers don’t get any ideas of world domination.