Terminator talk not just idle chatter

Editorial | 3 Aug 2017

We gasped when Google DeepMind's AlphaGo beat the world's best Go masters.

Then, many were just shocked when Facebook's chatterbots Alice and Bob invented a language of their own that's unintelligible even to their creators. The incident was, according to the Daily Telegraph, so dramatic that researchers were forced to shut down the pair of artificial intelligence robots.

These occurences seemed to have hit home scenes thought to exist only in movies.

The question being asked is: can the scenario of a world controlled by Skynet in the Terminator films become real one day - now that AI is proven to have the ability to beat humans, as in Go, and create a new language in the Alice-Bob mode?

Nobody can tell. But it's plain that scientists have been making enough progress in AI for Facebook giant Mark Zuckerberg to feel extremely excited, and for Tesla boss Elon Musk to urge caution.

That chatterbots Alice and Bob have created a language not intelligible to humans is of particular significance. It happened when they were paired in an experiment to imitate man's oldest form of trading: bartering.

Facebook insisted it was expected, and not a cause for alarm.

So, is AI good - or bad - for humans? It's an old question that has shadowed scientists since the first computer was invented. I fear that even if Zuckerberg and Musk are brought together for a face-tp-face debate, we still won't be able to find an answer.

Without a crystal ball, all they can say is to stick to their convictions. It's like an eternal battle of the heart and mind.

In a recent exchange between the tech titans, Musk mocked Zuckerberg for having "limited" understanding of the AI threat that Musk said could be a fundamental existential threat to human civilization. He also called for AI- related regulations.

The threshold that Musk - whose electric cars also use AI - has in mind could be the moment when robots begin to develop self-awareness. Perhaps, according to his probable baseline, that moment neared when Alice and Bob started communicating with each other in a language created on their own.

But could Musk be an alarmist? Zuckerberg would like to think so. Otherwise, he wouldn't have countered by calling Musk's warning "really negative" and "pretty irresponsible."

Zuckerberg's confidence is based on the premise that technology is neutral, and can be used for good and bad. To illustrate his point, the Facebook boss said AI could save lives by enhancing the safety of self-driving cars, and helping to diagnose medical conditions.

Zuckerberg was not wrong in his analogy. But Musk was on the mark too, for what he's concerned about is when robots develop consciousness - thus self-determination - and are able to ignore instructions from humans and make their own decisions.

The Musk-Zuckerberg debate is on a futuristic question. Yet, what had been viewed as futuristic in the past is now becoming real, as exemplified in the AlphaGo chess play and Facebook experiment.

Musk does raise a valid question: should there be checks and balances as humans continue to plunge full scale into AI?

Search Archive

Advanced Search
May 2018

Today's Standard

Yearly Magazine

Yearly Magazine