As a term, AI encompasses a range of supporting technologies including truly smart deep learning systems like IBM Watson and Google Deepmind; machine intelligence like the White House chatbot and intelligent assistants like Apple’s Siri.
While a machine as self-aware as HAL 9000 in 2001: A Space Odyssey may be some way off, we are clearly on a journey into the future.
We already rely on semi-smart machines for currency trading and presidential communications, but no AI is truly intelligent, even AlphaGo.
Neurala CEO Massimiliano Versace told Software Development Times: “Artificial intelligence is really taking the brain, and trying to emulate it in software. The brain is more than just recognizing an object. It is thinking. It is perceiving. It is action. It is emotion.”
It’s also about recognizing context and most AI aren’t good at this yet. Amazon, Samsung, Apple and Google all offer home automation ecosystems that can help accomplish relatively mundane tasks, such as changing the lights or the music. They aren’t true AI but use technologies that will support AI, such as voice recognition, pattern matching (to understand intent) and machine intelligence (to know the appropriate response).
The Presidential chatbot is another example of what is often called AI, but most bots (including the one from the White House) combine voice recognition with pattern matching, and there is usually a human in the machine to deal with questions the machine can’t answer.
Bots can convincingly match queries to pre-set answers, but aren’t yet truly smart. Even the 60,000 robots on the Foxconn production line lack the contextual reasoning to react to situations independently.
According to Gartner, “Smart machine technologies adapt their behavior based on experience, are not totally dependent on instructions from people (they learn on their own), and are able to come up with unanticipated results.”
The big names in tech are attempting to meet the AI challenge: Facebook recently open sourced its Big Sur AI; Google is developing cognitive machine intelligence, Tensor Flow; IBM continues to invest in Watson; Microsoft provides APIs for AI; and Apple is rapidly expanding its AI research, hiring Carnegie Mellon University professor Ruslan Salakhutdinov to lead its efforts and announcing an R&D center in Japan.
There are differences in the motivation of all these firms, but all seek to commercialize the technology at some point. For example, Facebook’s AI development represents the desire to show users precisely the ads they are most likely to respond to, based on the vast quantities of user data it collects.
Apple’s stated intention is to use machine intelligence to create useful user-focused product features to boost hardware sales, while others seek to create or identify new business models based on AI analysis of existing data sets.
There is great potential for good all the same. IBM and Germany’s Rhön-Klinikum’s Center for Undiagnosed and Rare Diseases are using Watson to analyze large quantities of complex medical data to help accelerate diagnosis.
And with the technology industry’s giants investing in AI, we could see rapid progress. Google director of engineering, Ray Kurzweil, believes we will have human-level AI by 2029. He says we are on a journey comprised of incremental innovations, each of which builds on those before to a greater understanding. “Technology goes beyond mere tool making,” he wrote. “It is a process of creating ever more powerful technology using the tools from the previous round of innovation.”
Facebook has taken this idea to heart and is developing an AI designed solely to build and test AI algorithms, enabling the social networking company to test over 300,000 machine-learning models each month. Successful models feed into the testing machine, enabling faster identification of other potentially successful models and more rapid tests on the journey to AI.
Mass-market sensor deployments, rapid proliferation of data sets and the evolution of supporting technologies such as voice, gesture or even emotion recognition are staging posts on the AI journey. All the same, CIOs would be wise not to buy too quickly into the hype: “If too many senior executives buy into anthropomorphic assumptions about conversational interfaces—for example, they are indistinguishable from people [or] they can learn by observing everything they need to know to replace all the people in your call center—then too many projects will fail and be shut down,” warned Gartner researcher, Tom Austin.