In recent years, the surge of artificial intelligence innovation and steady improvements to the associated technology has reached a point where the promises of the past can perhaps finally be fulfilled. A technology first introduced in 1955, AI has achieved many successes but also several failures, leading many to question whether we would ever witness its full potential for everyday use. Most people remember examples from the 1970’s and ‘80s when the HAL 9000, Star Trek computers and programs like ELIZA entered our collective consciousness. However, due to cost limitations and resource constraints, a lack of advanced technology and dwindling consumer interest, AI failed to realize those early promises and had retreated into research areas and highly specialized niches in the last 25 years.
Fast forward to today where we are once again in the midst of rapid AI innovation. Interest, capabilities and involvement in information technology are at an all-time high and the applications for AI have penetrated into thousands of everyday tasks. With the latest advancements in machine learning technology and growing consumer demand, we have recovered from the low point of AI and are looking towards seemingly endless possibilities. New-age voice-powered personal assistants like Alexa and Google Home, Apple’s HomeKit to control all electronic home devices, and the ubiquitous presence of chatbots streamlining processes from online banking to answering health related questions promise to grow the current technology wave and rebuild consumer trust and demand.
However, this new tide has led to one important question – is AI here to stay or are we merely in another bubble of unrealistic expectations that will burst a few years down the line? Although AI has penetrated everyday activities, is it sufficient to say that AI has managed to rectify all the misgivings of the past? Consider the recent spectacle we witnessed with the botched rollout and expectations of Tesla’s autopilot mode. Is the technology behind the modern AI movement – machine learning, big-data, data mining, deep learning, neural nets and natural language processing – worthy of the AI moniker?
The answer is both yes and no. Certainly, the recent advancements in artificial intelligence have proven to be of greater use and success in comparison to the past. While the past hype surrounding AI set unrealistic expectations for immediate consumer applications (given the limitations of the technology and data available at the time) advances in just the last five years have enabled AI to become a viable, mainstream business solution. Today’s technology thrives on data and the last five years have witnessed the accumulation of enormous amounts of data for this purpose. Coupled with improvements in database technology and increased computer horsepower to process the available data, AI has made a paradigm shift from scientific and academic use to widespread enterprise software consumption and consumer acceptance.
The success of today’s AI movement is largely the by-product of a few critical factors including new platforms from major players in the field including Google, Microsoft and Amazon, which have supported the mainstream practice of AI, building a critical mass of practitioners leveraging these platforms. Further, commitments from large corporations and technology leaders including IBM, Yahoo!, Salesforce and Apple have helped secure a place for AI in the future as they double down on the technology to improve processes in areas such as data security, computer-assisted diagnosis in healthcare, purchase prediction, fraud detection and much more. Long-term investments by these big players are further evidence toward the staying power of AI this time around. If you combine a lower price point for the technology, commitments from industry giants, a growing mass of available data waiting to be analyzed and changing consumer expectations about what is achievable, it becomes clear that the renewed interest in AI is real and not a bubble that is ready to pop.
But are these companies leveraging genuine AI technology? As the performance of AI is at an all-time high, one must understand that “artificial intelligence” in its truest sense is still years, if not decades away, as machines have yet to operate independently of human intervention. Technology like machine learning, data mining and neural nets – while all classified under the AI moniker and driving the next generation of solutions – cannot learn on its own or invent anything new with their knowledge. As ironic as the name AI implies, machines cannot adequately function without relying on known data sets and pre-programed responses and behaviors. In other words, AI can find patterns in enormous volumes of data – even patterns that a human would fail to see – and rapidly and efficiently handle routine tasks, yet they cannot invent novel solutions to problems due to their dependency on programmed algorithms.
AI as we know it today leverages modern advances in statistical techniques that some may say do not qualify as traditional AI and are not inherently ‘intelligent’ however this recent resurgence of AI has proven to be very useful across a variety of applications for both business and consumer. How far we advance AI depends on the continued investment of big industry players, additional improvements in data techniques and algorithms and applications that consumers and businesses find useful.
by Terence Davis – Terence holds a BS in Astrophysics from UCLA and formerly served as VP-Product Management for an SDN/NFV startup delivering solutions for US and European customers.