Oct 16, 2017

Artificial Intelligence Blog Series - VOL 1

Vol 1 – is Pinocchio a real boy yet?

SPOILERS - We already know that our favourite wooden puppet, with the help of Jiminy was able to achieve his dream of becoming a real boy. What you might have forgotten is that there are a few important caveats to the story – in order to become a real boy Pinocchio needed to first prove himself to be ‘brave, truthful, and unselfish’ by accepting responsibility and being honest.

Taking a second look, it was surprising to me that the classic fairy tale shares a number of consistent themes with more modern sci-fi films like ‘A.I. Artificial Intelligence’ and ‘Ex Machina’. Whether it was Pinocchio, David, or Ava, each character was brought to life and given an opportunity to embark on a journey of self-discovery. The biggest difference however, was that Pinocchio was the only one to follow the path designed for him.

For almost eighty years these films have opened our minds and shaped our thinking around the potential impact of Artificial Intelligence (AI). While each film offers an exciting portrayal of a not too distant future, we still don’t have a specific timeframe for when AI will make the transition from the movie reel to a practical application in our daily lives.

There was a similar period of doubt back in the 1960’s when the Soviets and Americans were competing to be the first nation to put a man on the moon. Uncertain about the outcome, it was only when President John F. Kennedy said the following that many people gained confidence in their ability to reach their goal.

"I believe that this nation should commit itself to achieving the goal, before this decade is out, of landing a man on the Moon and returning him safely to the Earth. No single space project...will be more exciting, or more impressive to mankind, or more important...and none will be so difficult or expensive to accomplish..."

At that time, these words had inspired a transformative change in the ideological, technological, and political aspirations of two rival nations – and more strikingly, a shift from what was considered hype to reality.

I believe that we are at the cusp of a modern day ‘Space Race’ and as the explosion of data borders immeasurable and advances in technology hit exponential rates, we are presented with an opportunity for AI to become the next big event of our lifetime.

Modern examples for cognitive solutions like Apple’s SIRI and Amazon’s Alexa have already illustrated a role for narrow AI in our everyday lives and it is only a matter of time until we begin to see examples of general AI technologies like the characters for ‘Ava’ and ‘David’ mentioned above.

In the last year alone, we have seen significant investments made to increase the development of AI applications, with Deloitte reporting $3.1 billion worth of global M&A deals last year, and another $1.2 billion in corporate venture investments.

As the market continues to grow it’s reasonable to assume the development of AI applications will follow the trend, and as we work our way towards an ‘intelligence explosion’ it might even be here sooner than we think. Futurist Ray Kurzweil is already predicting that computers will have human-level intelligence by 2029.

The main challenge facing us is whether organizations are ready to adopt these applications and technologies. For instance, we can look at a recent example from Facebook. What the Silicon Valley giant did was train its bots to negotiate with one another. The problem was – and this was the part that made international headlines – their bots went ‘rogue’ by developing their own ‘non-human’ language (aka Facebotlish) and as a result, the team shut down the program and had to go back to the drawing board.

Amongst the growing list of challenges facing AI, one of the biggest hurdles the industry is facing originated in January 2015 when an open letter was released by dozens of experts (including Stephen Hawking and Elon Musk).

The letter acknowledges that “potential benefits are huge, since everything that civilization has to offer is a product of human intelligence”, but ultimately stresses that “because of the great potential of AI, it is important to research how to reap its benefits while avoiding potential pitfalls.

One of the most controversial examples of this has been the Trolley Problem for the autonomous vehicle market – what if a fully autonomous car is driving down a highway and the brakes malfunction, does it continue to drive down the road and hit the 5 pedestrians standing on the road, or veer to the left side of the road in front of a speeding Mack truck? - Check out this episode of Radiolab for more.

Similarly, Nick Bostrom provides a viewpoint on the long-term pitfalls of AI (popularized through his book ‘Superintelligence: Paths, Dangers, Strategies’ and his TED talk). Focusing on the implications and consequences of a more advanced machine superintelligence, he provides a number of examples for general artificial intelligence and explains that the creation of the first superintelligent agent will be the “the last invention humanity will ever need to make”. While Nick is confident that these machines will have the capabilities to solve our world’s toughest challenges, he expresses a concern around what happens when computers become smarter than we are and how they will deal with the trolley problem listed above.

While it is important to think about the philosophical side of AI, it is clear that we are still decades away from truly being able to understand the implications of a machine superintelligence. Adam Cheyer, co-founder of SIRI, summed it up well at a recent event in the Cayman’s:

“In my view, there is nothing to worry about. If you talk to anyone who works in AI and understands what AI can or cannot do, it will be more likely that we will meet extraterrestrial life than machine life.”

Cheyer continued, stating that AI is very good at solving narrow problems where neural network-trained machines are capable of beating the best human players, like self-driving a car or playing chess.

“But we have made almost no progress in general AI. Any 5-year-old is way smarter than any machine in extracting core ideas from one situation and applying them to a completely different situation. We do not know how to do that. Consciousness, we have no idea what that means. There is no research that is making progress into that.”

I feel quite optimistic around the benefits AI applications can offer, but as we continue to see a number of prominent researchers and scientists offer different viewpoints on the subject, it is worth pausing to wonder whether we are taking the right approach towards developing these smart machines. The big question I keep asking is whether future applications of AI will impact our society by helping preserve humanity and our values – or whether they will have values of their own.

It’s safe to say we can begin to expect that AI applications will transcend the movie screen and begin playing a key role in our society – but before they do we should remember our friend Pinocchio and start questioning whether being brave, trustful, and unselfish will be enough.