When IBM launched its Watson cognitive computing project over a decade ago, it started as an effort deep in IBM’s research labs to produce a system that could learn and think like humans.
Since then, Watson proved its mettle by winning the TV game show Jeopardy in 2011 and spurring interest in cognitive technology across the industry.
IBM is not alone ushering in what Big Blue’s CEO Ginni Rometty has dubbed the “cognitive era.” Several other companies have prominent cognitive technology offerings or projects in the works, including Amazon, Apple, Facebook, Google, HPE, Microsoft and SAS, among others.
Driving this new era are data—particularly unstructured data—and cognitive systems that can ingest all types of data and work in natural language.
Cognitive systems can understand, reason and learn, Rometty said. In short, cognitive computing is the use of systems that can think and learn.
Indeed, cognitive computing encompasses various forms of artificial intelligence (AI), including machine learning, reasoning, natural language processing, speech and vision, human-computer interaction, dialog and narrative generation and more. However, cognitive computing is but a subset of AI.
Many attempts have been made over the years to deliver cognitive systems, but until recently these systems haven’t been very prevalent or cost-effective. They are more cost effective now with the advent of low-cost, high-performance computing power and data storage.
Now businesses are increasingly looking to cognitive technology to help cull through reams of data to glean insights on everything from customer buying patterns, to medical diagnoses and treatment options, to financial advice. As demand for cognitive assistance grows, so has developer interest in building cognitive applications.
Cognitive Banking and Machine Learning
Capital One, which has emerged as a forward-looking, developer-centric financial services institution, is working on delivering cognitive applications to help its customers.
“Being a leader in this new world of banking is going to require competencies around the ability to innovate quickly, the ability to build software that is really compelling and well-designed and [has] the ability to understand data and analytics,” said Robert Alexander, CIO of Capital One.
The financial industry will “apply modern methods of machine learning and analytics operating in real-time in a way that banking just doesn’t know how to do,” Alexander said.
Capital One is applying cognitive computing in a number of different areas, Alexander said, particularly in cyber-security. Capital One is using machine learning to help secure systems, starting with building a comprehensive data infrastructure of what’s going on in their computing environment and building tools that offer insights into things that might be anomalies, he said.
“You can use machine learning and AI to discriminate normal behavior from abnormal behavior,” Alexander said. “So we’ve applied machine learning in that context to help us identify malware or anomalous log-in behavior or other kinds of things that are indicators of threats in our environment. Other places are around fraud and risk decisions.”
More and more, algorithms are taking over things that humans used to do, said Alex Backer, CEO of QLess, a startup that produces an app that uses AI and machine learning that aims to eliminate the need for people to wait in line at stores, restaurants, government offices and more.
“The idea for QLess came from me standing in line and thinking that there ought to be a better way to get service than standing behind other people’s butts,” Backer said.
Companies Large, Small Exploring Cognitive Technology in Latest Apps
“It allows people to join a virtual mobile line from any mobile device and in the period that they’re waiting they get notified of how close their turn is.”
As a cognitive computing system, QLess and its algorithms are self-correcting—like GPS systems, because they continue to learn over time, Backer said. To build the QLess app, Backer and his team built a lot of their own custom software.
Microsoft Goes Cognitive
Meanwhile, at Microsoft’s Build 2016 conference, the company introduced Microsoft Cognitive Services, a collection of knowledge APIs that enable developers to make their applications more intelligent.
It includes APIs that enable systems to see, hear, speak, understand and interpret data using natural language. With Cognitive Services, developers can add intelligent features—such as emotion and sentiment detection, vision and speech recognition, knowledge, search and language understanding—into their applications.
At Build, Microsoft also introduced its Cortana Intelligence Suite, formerly known as the Cortana Analytics Suite. Both Microsoft Cognitive services and the Microsoft Bot Framework are part of the Cortana Intelligence Suite.
“We want to give you a set of micro services, these cognitive services, so that you can have language understanding, speech understanding, computer vision built into your applications, and also rich machine learning capability, because we think of this intelligence run time and Cortana Intelligence Suite” as going to be the core of the latest applications, Microsoft CEO, Satya Nadella, said in his keynote at the Build conference.
Is IBM Exaggerating?
Still, despite a host of players competing in the cognitive systems arena, IBM’s Watson remains among the most prominent entries and IBM is aggressively courting developers to adopt its cognitive platform via the Watson Developer Cloud.
However, even as Watson holds the largest mindshare and marketing share in the field of cognitive systems, it is not without its detractors. In a recent article, Roger Schank, an AI theorist and CEO of Socratic Arts, took IBM to task for its Watson claims.
“Watson is not reasoning,” Schank said in his post. “You can only reason if you have goals, plans, ways of attaining them, a comprehension of the beliefs that others may have, and a knowledge of past experiences to reason from.”
In the post entitled The Fraudulent Claims Made by IBM About Watson and AI, Schank added: “They are not doing ‘cognitive computing’ no matter how many times they say they are.”
Elliot Turner, director of Alchemy in IBM’s Watson division, brushed Schank’s criticism aside. “One of the things I always found about the world of cognitive technology and AI is that if you ask the folks in the research or academic community, they’ll say AI is whatever is not working yet,” Turner told eWEEK.
It’s a scenario where the goalposts are always moving, he said.
“There was a time when people thought AI was the ability for a computer to understand human speech,” noted Turner. “Now we’ve reduced that to a technology we call speech recognition.” Turner was founder and CEO of AlchemyAPI, which IBM acquired last year.
“I think there’s a natural and reasonable inclination to doubt vendors’ claims about their progress but the attacks by some on IBM and Watson have been over the top,” said Charles King, principal analyst at Pund-IT. “In certain cases, they’re arising from companies and individuals whose own efforts in cognitive technologies are being overshadowed.”
Companies Large, Small Exploring Cognitive Technology in Latest Apps
Among the many companies relying on Watson for their own AI operations is Equals 3, which uses Watson to power Lucy—its cognitive companion for marketing professionals. Lucy uses several IBM Watson services, including Retrieve and Rank, Natural Language Classifier, Personality Insights and Tradeoff Analytics. Lucy can comprehend vast amounts of structured and unstructured marketing data and provide detailed answers in seconds.
“Lucy solves the data-overload problem marketers and agencies have been grappling with for years: accessing the appropriate data at the right time,” said Scott Litman, co-founder of Equals 3, in a statement.
“The fact that Lucy can accomplish in one day what has previously taken months to do is a complete paradigm shift, and her ability to enhance what marketers can do is really made possible thanks to the cognitive computing capabilities of IBM Watson,” Litman said.
Watson Not for Everybody
However, despite there being more than 80,000 developers using the Watson Developer Cloud, IBM’s cognitive system is not a match for everybody.
Rob May, CEO of Talla, which markets an AI-powered virtual assistant akin to Apple’s Siri, but aimed at business professionals, said Watson was simply not a fit for Talla.
“We do a lot of natural language processing and we have a cognitive computing approach to a lot of what we’re doing,” May said. “We tried Watson in the beginning and it really just wasn’t a very good fit for us.”
May said there are a lot of pros and cons to Watson. Among the pros are that IBM is clearly committed to it, the Watson resources will be available long term and they offer a lot of different functionality.
But on the down side, said May, a lot of Watson “is maybe more marketing hype than substance in some ways. It definitely does some stuff, but some of the functionality is pretty limited and it’s really not set up well for small companies to build products on top of. It seemed like a lot of the Watson functionalities that are set up are targeted at larger IBM customers, which would make sense.”
In particular, he believes Watson seems targeted at existing customers that have large data stores that are appropriate for cognitive computing, or some kind of analysis that can’t be readily performed with common, existing algorithms.
Rather than use Watson, Talla used a series of open-source components and wrote a lot of its own cognitive code.
“IBM has tried to get companies like us to build on top of their platform, but I just don’t think it’s ready yet,” May said. “It’s not flexible enough; it’s not easy enough to get started.”
Yet, IBM has sponsored a number of Watson-related hackathons and developer competitions where small startups have successfully used the services to launch new apps.
Moreover, IBM’s recent deal with Twilio to integrate Watson services with the Twilio communications platform not only opens Watson up to the more than one million developers in Twilio’s developer community, but it only takes two minutes to integrate the cognitive services with Twilio, said Patrick Malatack, vice president of product management at Twilio.
Right now, IBM is doing more than its competitors to encourage developers to learn how to create cognitive applications, Pund-IT’s King said.
“I can’t think of another vendor among IBM’s direct competitors that is supporting similar developer-oriented efforts to anything like the same degree,” said King.
“That said, companies including Apple, Google and Microsoft have created powerful solutions supporting natural speech and machine learning functions. Google and Microsoft also have proactive, innovative efforts aimed at increasing developer activities around those platforms.”