THE BLOG

Artificial Intelligence: No Imitations, Some Limitations

17/12/2014 04:41 GMT | Updated 15/02/2015 10:59 GMT

In this digital age, concepts such as Artificial Intelligence (AI) are impacting us more than ever before. AI is now featuring heavily in the news and in cinema, most recently in historical drama and biopic, The Imitation Game, which recounts the life of Alan Turing, his mathematical contributions to code-breaking and the winning of World War II. Turing was the first to question whether machines could be engineered to think autonomously and proposed The Turing Test to measure an artificially intelligent system by pitting it against a human. In his report from 1950, Computing Machinery and Intelligence, he invents a game to be played by a human, a machine, and an interrogator to find out if machines can think. If the program can convince the interrogator that it's a human, it passes the test and has reached the intellectual capacity of a human.

The Turing test has long been regarded as the primary means of measuring human-level intelligence in a machine however modern day scientists have been looking at an alternative barometer for this, called the Winograd Schema Challenge. This approach was proposed by Hector Levesque at the University of Toronto as a more accurate means of measuring a machine's reasoning abilities. While the Turing Test is based on free-form conversation, the Winograd Schema Challenge gives a set of multiple-choice questions that evaluate a human or machine's common knowledge of the world. For example:

Q The trophy would not fit in the brown suitcase because it was too big. What was too big?

Answer 0: the trophy?

Answer 1: the suitcase?

A human typically would use their general knowledge on object size to determine the correct answer, however most AI systems currently lack the reasoning needed to deduce that conclusion. With gathering consumer interest in AI, expectations should be managed to reflect current technological limitations.

Last year's sci-fi movie hit HER directed by Spike Jonze, focused on the relationship between a man and his operating system Samantha who was personified by a human female voice. Samantha could engage in conversation, feel emotion and make sentient choices which, given our earlier example, far exceeds reality! What is currently possible however, are virtual personal assistants that can interact with us in natural ways through Natural Language Processing (NLP), simplifying our daily lives through smart phones or in-car technology, from scheduling appointments to updating our Facebook status and making calls. The virtual personal assistants of the future will track context, extract meaning, and embody individual personalities that will lead to a long-awaited humanised technological landscape. This contextual awareness coupled with automated cognitive reasoning is leading to intelligent systems whose actions will have a greater influence in people's lives, leading to personal virtual assistants that can behave in a more proactive and anticipatory fashion.

Additionally, the idea of associating an intelligent assistant with a single device is already a thing of the past; instead, assistants are already manifesting themselves through a variety of hardware that we utilise throughout the day. These AI-driven virtual assistants will simplify the often-overwhelming spectrum of content, services and capabilities that we have access to through phones, PCs, tablets, TVs, cars, apps - and now watches, thermostats, and an expanding array of consumer electronics. Moving forward, to render these devices even more capable and intelligent, it's critical that these attributes are matched by the ability to reason and understand what you and I want, need and expect out of them.

Stephen Hawking* recently commented that AI 'could spell the end of the human race' querying the power of existing technologies such as Apple's Siri app and whether machines could eventually overtake humans. Undoubtedly these comments shine the spotlight on AI ever more brightly, bringing certain scenes from The Terminator into focus. But in our mind, it has a long way to go for that prediction to ever be a reality. Fears about AI are based on the premise that as species become more intelligent, they evolve to be more controlling and violent, and we believe the opposite to be true. As we become more intelligent, as a race we become kinder and treat people better. But ultimately, we envision AI not existing to create artificial companions, but to assist and amplify our own capabilities through the creation of transformative tools. Whilst this may not be in the immediate future, it's not too far off, for which we have the likes of Alan Turing to thank, bringing renewed emphasis onto this exciting journey we are on. Happy watching.

Notes:

* http://www.huffingtonpost.co.uk/2014/12/02/stephen-hawking-ai-end-of-the-human-race_n_6254010.html