From simple games to self-learning systems – AI has come a long way
By Ravi Chalaka, Vice President, Global IoT and Lumada Marketing, Hitachi Vantara
Today, talk of artificial intelligence (AI) is everywhere – from Apple’s Siri to how Uber dispatches drivers, to the way Facebook arranges its Newsfeed. However, it wasn’t long ago that AI was regarded by many as purely science fiction – the plotline of Hollywood blockbusters, such as 2001: A Space Odyssey. The term “artificial intelligence” was coined in 1956 with the hope of creating machines that could emulate human intelligence, such as reasoning and judgement.
Those early days drew scientists from academia to enterprise, to ignite a revolution of innovation that we’ve been riding ever since. Here are three significant waves of AI that have brought us to where we are today – while the first two generated hype and no real commercial adoption, the third time was the charm, which has already produced many real world use cases, taking us closer to the fulfillment of that early vision.
First Wave: Age of Search and Deductive Reasoning
The first wave of AI generated a lot of interesting ideas in the 1950s. Engineers devised deductive reasoning programs with logical rules, a method in which a conclusion is based on multiple premises that are assumed to be true. This approach led to simple games such as the first computer program capable of playing the game of Draughts. Although these first-wave AI systems could perform straightforward reasoning tasks, they were unable to learn anything on their own. While they couldn’t be applied to business at the time, some of today’s applications in smartphone maps can be traced back to this first-wave of AI technologies. But in general, this wave did not have any material impact on business or people’s lives.
Second Wave: Age of Knowledge Acquisition
The second wave of AI began in the 1980s with big AI projects when researchers turned their focus to helping machines acquire knowledge. Instead of just programming precise rules for machines to follow, they tried to teach machines the knowledge of experts, and developed statistical models which machines could use to adapt this knowledge to different situations. While second-wave AI machines had some breakthroughs – Deep Blue became the first computer system to defeat a reigning world champion Gary Kasparov in 1997 in chess – many of these second-wave expert systems struggled with accuracy due to complexity issues, rendering them impractical for most business applications. Commercialization of AI was still elusive.
Third Wave: Age of Machine Learning
Since the turn of century, we’ve been riding the third wave of AI, where computers use machine learning and deep learning techniques to automatically learn from vast amounts of data and improve from experience without being explicitly programmed with knowledge. Immense processing power enabled development of artificial neural networks, better natural language processing, and enhanced image processing. Third-wave AI is now able to consume data from statistical models, identify patterns in the data, create common sense rules, and incorporate information from multiple sources to reach a conclusion on their own. For example, AI – along with internet of things (IoT) connectivity – is extracting data from wearable devices and public sources to create personal health updates in real time. We’re seeing third-wave AI driving numerous commercial applications in personal assistants, such as Amazon’s Alexa and operating self-driving vehicles used in industry. This wave is expected to last for a long time while dramatically changing the way we live.
While the journey from the simple games of the 50s to the self-learning machines of today has been quite a ride – we have just scratched the surface of the full potential of AI. One thing is for certain: AI is more than Hollywood storytelling. It has become essential for suggesting alternate driving routes in rush hour traffic, making online shopping recommendations – even helping airport security officials use AI with video cameras to perform live face matching to recognize criminals within a crowd at busy terminals.
But the big game changer is applying machine learning in industrial environments, such as manufacturing, transportation, mining, agriculture, and energy to predict failures, recommend options and even augment human activity to build safer cars, sustain the planet, achieve mass customization, and reduce waste. Hitachi is making significant investments in AI’s machine learning and deep learning research and development to deliver solutions for the industrial world with its AI-enabled Lumada IoT Platform.
In a future blog, I will discuss the future waves of AI, when AI becomes ASI (Artificial Super Intelligence) which means it first catches up and subsequently surpasses human intelligence. Don’t worry we are many, many decades away from that wave.