Machine learning: The next frontier for intelligent analytics solutions
Machine learning: The next frontier for intelligent analytics solutions
Progression of technology
Technology has advanced to the extent that the sci-fi stories have come close to becoming reality. Whether it is the humanoid AI from “Ex-Machina,” Skynet from “Terminator,” or JARVIS from “Iron Man,” most people might likely agree that nothing seems impossible to achieve. The debate lies on how far are we from getting there. And here’s why we are probably much closer than most of us might actually think.
Ray Kurzweil – the American author, inventor, futurist well known for his predictions about artificial intelligence and the human race – suggests in what he calls “the law of accelerating returns” that the rate of change in a wide variety of evolutionary systems tends to increase exponentially. This includes the evolution of technology as well. Kurzweil suggests that this exponential technological growth is counter-intuitive to the way our brains perceive the world, as our brains were biologically inherited from humans living in a world that was linear and local. Due this exponential nature of growth, all predictions made based on past and present growth rates would lead to massive underestimations of the future, which in turn would lead to great skepticism in our future projections. If Kurzweil is correct, the level of advancements we would experience by jumping to 2035 would be equivalent to what a person from 1750 would experience in 2016.
So, advances are getting bigger and happening more quickly than before. This suggests some pretty dramatic things about our future, right?
This theory is well demonstrated in the case of business services (business process and IT services) in terms of the adoption of Robotic Process Automation (RPA) – or rather, the lack of it. The role of technology (RPA in this case) in delivering services has evolved at a very fast rate, faster than what one would have naturally perceived, and the skepticism in future projections resulted in most enterprises’ seeming unpreparedness for robotic automation. The economic downturn that started toward the end of the last decade should have been the perfect trigger for RPA uptake in business services. RPA adoption seemed like a natural corollary as enterprises concentrated feverishly on cost savings, and trimmed their support functions in the wake of the recession … but that didn’t happen. A very important technological development has been presenting itself in the form of RPA, and most enterprises were clueless about the impending disruptions. It was not until recently that enterprises started to take a serious look at RPA. Even now, action is lagging the hype, though the upward trajectory for RPA adoption from here onward should be exponential.
Era of Cognitive Disruption – the road to Artificial Intelligence
Cognitive disruption and its usage in business services is an extension of RPA’s story. Many people are confused about the term Artificial Intelligence, mainly because it’s a very broad subject with diverse set of applications ranging from smart phone apps to self-driving cars to something much more dramatic in the future, and because of the way it’s portrayed in popular media.
Almost all the AI and cognitive platforms that have been developed to date, ranging from iPhone’s Siri to Google’s self-driving cars to more sophisticated systems such as IBM Watson and IPSoft Amelia, are examples of what are called Artificial Narrow Intelligence (ANI) systems – AI that equals or exceeds human intelligence in specific areas. The next generation of AI, Artificial General Intelligence (AGI) – AI that is as smart as a human and can perform any intellectual task that a human being can – may still sound like science fiction; but it could suddenly become very real due to rapid advancements in technology. We keep coming across claims made by various AI developers about how close they are to achieving the next level of cognitive intelligence. For example, IPSoft claims that its artificial intelligence platform, Amelia, is close to achieving “near human cognitive capabilities” and we are going to hear more and more about AI in the coming months and years. However, many such claims are still met with great skepticism, and understandably so.
AI-enabled automation of knowledge work could cut employment costs by $9 trillion by 2020, according to estimates by Bank of America. This depicts the huge potential for near-complete automation of core and repetitive businesses functions in the future. Many enterprises missed the early adopter bus for RPA, in a perfect example of the law of accelerated returns. The question is, “Will they repeat the same mistake for AI?” Or, more correctly, “Are they ALREADY REPEATING the same mistake for AI”?
Being reactive is certainly not the best way to realize full benefits of important technological advancements, especially in highly competitive markets. In order to be future ready, enterprises need to cut through the web of skepticism, and proactively take necessary steps to align themselves to what is about to come. Due to the counter-intuitive nature of the progression, this is probably even more challenging than it actually sounds.
Business services is inherently an area which is seemingly laggard in any types of innovation. Most enterprises are reluctant to doing untested innovation in back-office processes. The primary challenge is the chicken-and-egg problem… getting budget allocation without demonstrable benefits out of using AI, and vice-versa. Enterprises can move past this problem by starting small with a small seed budget, creating liquidity in a small segment of the market, getting the virtuous circle to work for them, demonstrating some benefits, getting stronger buy-in, obtaining a bigger budget for bigger AI projects, and extending to adjacent areas.
Several companies, such as Baidu, Black Knight, Facebook, Google, IBM, Hitachi, and Microsoft are investing heavily in AI. And they are not only at the forefront of the latest technological developments, but are also laying the groundwork for the future developments.
Technological advancement is progressive, and organizations need to prepare for this journey, rather than seeing it as a destination. Enterprises are already in the midst of robotic revolution, and with Kurzweil’s law of accelerating returns in mind, it is time they start embracing the cognitive era.
Last week, Wipro’s CTO briefed me on the Wipro Holmes artificial intelligence platform. My key takeaways from the session and subsequent musings on where AI is taking the industry:
Overall, the AI platform market is focusing on three broad areas:
As of now, managed service providers like Wipro focused on the first two – and understandably so. Innovation using AI is seen in the context of the broader business model and differentiation in their core markets rather than risky investments in areas that are not fully understood – yet.
All of this might change. As the old aphorism goes, we tend to overestimate the short term and underestimate the long term. As the world goes increasingly digital and different business models involving a nexus of technology and service providers, user and developer communities, and adjacent industry participation come to the fore, it may not be long before services providers realize that it’s a question of “and,” not “or.”
One prediction I have made about the future of service delivery automation (SDA) is that increasingly enterprise software will have the technology embedded. This is particularly true of intelligent and cognitive type of tools. I expect these to become a common feature of enterprise software in the next 5-7 years.
We saw this kind of trend in the earlier days of business intelligence and reporting. The popularity of third-party tools saw the functionality built into enterprise software. As well as reports on activities, dashboards started to feature in applications giving instant views of what was going on in the enterprise. We do not have to look far to find such software today, for example, Blue Prism, includes analytics that report on operations and performance of its robots.
A current example of a more intelligent enterprise software is Oracle Policy Automation Cloud Service. This reads policies written in natural language. Then based on business rules and the policy, it decides what questions to ask the customer, performs eligibility checks, and produces a decision report.
Another example is HighSpot, an enterprise search tool that uses natural language processing for searches and machine learning for finding the most relevant information and ranking the results.
The availability of open source machine learning software libraries, such as Apache Mahout, and software tools from industry giants, such as Microsoft (Machine Learning Service on Azure), will accelerate the next generation of smarter enterprise software.
Some would say that intelligent enterprise software would be function-specific, but I believe some varieties will be able to do more than one thing within large software applications. The need for standardization of interfaces to these tools and the ability to interact with other intelligent applications will grow over time too. We could even see more automations crossing paths across workflows leading to more complex machine-based decision making.
The question is what impact will pervasive intelligence have on the outsourcing industry:
Intelligent enterprise software is here. And we are on the brink of it becoming pervasive and commonplace. As it does, I’ll continue to share my insights on its evolution.
Photo credit: Flickr
The recently announced acquisition of Merge is one in a string of initiatives by IBM to increase both its market presence and depth of offerings to the healthcare sector. With birth rates increasing in many parts of the world and the aging population growing in developed countries, the race is on for data driven and highly efficient healthcare.
IBM is clearly targeting this market. Its recent activities have included:
The US$1 billion acquisition of Merge brings IBM a medical imaging platform to combine with Watson’s image data and analytics capabilities and an extended client base. Excellent and Elementary, Dr. Watson.
With these initiatives, IBM is building specialist competences, to capture, analyze, and recommend treatments or actions that would help healthcare providers, payers, pharmaceuticals, as well as individuals achieve positive health outcomes.
Gaining a wide range of capabilities in specific areas has helped IBM generate specific segment revenue in good and bad times. For example, its large number of information management and WebSphere portfolio acquisitions (e.g., Cognos, Netezza, and SPSS, to name but a few) has seen segment-specific revenues maintain steady growth over the years.
If IBM was to successfully combine its deep specialization in healthcare with Watson’s cognitive computing to enhance its services, it could gain a big edge over competitors at a time when demand is set to grow. At the moment we are seeing more of IBM in healthcare IT infrastructure modernization contracts than data-driven care provisioning and support services. Recent examples include:
These types of contracts give IBM opportunities to tap into new solution and services openings at existing clients.
Other challenges for IBM’s intelligent and data driven healthcare offerings include:
IBM is going all out when it comes to showcasing Watson as a competitive differentiator. In an uncharacteristic move (and a sign of the times), it has launched Watson Developer Cloud, an open platform for developers to build apps on top of Watson for industry-specific solutions (through a set of APIs and SDKs). It is also working with app developers such as Decibel, Epic, Fluid, Go Moment, MD Buyline, TalkSpace, and Welltok to build apps embedded on Watson technology, thereby, rounding up a robust ecosystem. It is abundantly clear that IBM views healthcare as the principal vertical where Watson’s computing prowess can make its mark. In the meantime other service providers are likely to build or acquire their own cognitive capabilities to challenge IBM on pricing and specialist offerings.
Photo credit: Flickr
I recently spent a very productive day with Wipro as they showed examples of their commitment to service delivery automation – a commitment I observed as more than in any other service provider. HOLMES, their recently unveiled artificial intelligence platform is just the beginning of this serious commitment. Here are three very important aspects of this commitment.
Some time ago Wipro recognized that the service industry is changing very profoundly, and a significant secular driver of the change is new automation technologies that allow customers to automate a dramatically higher number of services than has been automated in the past. They took concrete steps to address this change.
They recognized not only the customer impact of implement automation in their workflow but also the impact to the service provider. Over time, the value will be captured by the automation owner, not by the service provider. Therefore, Wipro decided to invest in owning its automation IP – and that involves not just funding the initial build but also ongoing investments.
That’s not to say that when Wipro deploys automation in their accounts they will employ exclusively their own IP, but they recognize the need for a significant portion being their own IP.
Wipro allocated a large, dedicated team into building its automation platform, which they named HOLMES. But unlike some other providers, they extensively used open source software to get a head start and then layered in their own development on top of the open source component. This allowed them to move quickly in bringing compelling functionality to the marketplace.
The third aspect of Wipro’s automation strategy is their commitment from CEO TK Kurien on down through the leadership ranks to bring this to the marketplace. TK and senior leadership are committed to take this service delivery capability into their existing client base as well as use the automation platform as a challenger to gain new share in the market.
I believe bringing automation into their existing client base will be the most challenging endeavor, and they acknowledge that it will be disruptive and may be cannibalistic to their revenue flows.
We await to see how they handle such disruption. The details revealed to me were somewhat vague as to how they will realign their incentives to allow their account teams to do this. But certainly the executive commitment is such that it’s possible they will take the necessary steps to make incentive changes.
They are preparing for the inevitable disruption that will accompany the drive to become a leader in automated service delivery.
The new technologies sweeping the market hold great promise of competitive advantages. But there’s a disturbing trend occurring in the services sales process for these technologies that poses a risk for buyers. Look out for providers talking about cloud, mobility, big data, the Internet of Things, and social in the same breath as SaaS/BPaas, automation, robotics, and artificial intelligence. Providers that jumble these technologies together as though they are homogeneous really don’t understand the implications of what they’re trying to sell you. They’re basically throwing mud against your wall and seeing what sticks.
The possibilities with all of these technologies are exciting, but they have distinctly different impacts on the buyer’s business.
As illustrated in the diagram below, we can bucket one class of impacts as those that create new business opportunities. They provide new types of services that enterprises can use to change the composition of their customers or provide different kinds of services. For example, the Internet of Things holds enormous promise around allowing enterprises to provide a completely different class of services to their customers. In mobility and social technologies, the digital revolution holds the promise of changing the way businesses interact with their end customers.
The second class of new technologies (Saas/BPaaS, automation, robotics, and artificial intelligence) changes how services are delivered. For example, SaaS takes a functionality that was available but delivers it through a different mechanism. Automation and robotics changes the way service is provided by shifting from FTE-based models into an automated machine-based delivery vehicle.
The two buckets of technologies have different value propositions. The first class of technologies (cloud, mobility, big data, IoT, and social) are about getting new and different functionality. The impacts in the second class are lower costs and improved flexibility and agility. Each class of technologies has different objectives and value propositions and thus needs a different kind of business case. Buyers that mix these technologies together in a business case do themselves substantial disservice.
The way you need to evaluate the two distinct types of technologies (and providers offering them) is completely different. A provider that recognizes that automation, robotics, and SaaS are about changing the nature of delivery will have a much more thoughtful conversation with you and build its value proposition around flexibility, speed, and quality of service and cost.
A provider that recognizes the impact of mobility, cloud, big data, and the IoT technologies will talk to you about a value proposition around standing up exciting new capabilities, creating new offers and changing the conversation with your end customers.
So, buyer beware. If you’re talking with a provider that mixes these technologies’ distinct value propositions together, you’re dealing with a provider that really doesn’t understand what they’re offering.
Photo credit: Flickr
I had the privilege of being at IBM and seeing first hand Watson working on powerful use cases. I must say, even now after a few days of reflecting on it, I think I’m even more impressed with its power and capability than when I was at IBM and saw Watson in use. If, like me, you spend two hours with Watson, you will get a glimpse at our future. It’s highly likely that within five to 10 years all of us will use some kind of cognitive computing to assist us in our daily lives. But I believe there is a major challenge.
Just a quick refresh: Watson is cognitive computing, a form of artificial intelligence. Previously I did not understand the way it will be deployed; it will augment human decision making, not replace people. That’s not to say that an individual assisted by Watson won’t be able to do the work of many more individuals. At least at this stage of development cognitive computing makes humans more capable and smarter.
For example, I saw Watson working as a companion to an oncology doctor, helping him perform more thorough diagnostics. In the situation I observed, the oncologist was able to cut the diagnosis and testing process from six days down to two hours. That doctor was far more effective because Watson can explore many more options and present hypotheses and data to the doctor and medical team than they could have explored on their own (plus it would have taken far more time for them to do it). In addition, it’s not hard to believe that the team would be more likely to do a better diagnostic with Watson as companion than they could achieve through traditional techniques.
With all that being said, I think Big Blue faces a major challenge with Watson at the moment: Watson is a solution looking for a problem.
As I understand it, IBM invested over a billion dollars in Watson’s development. On TV we saw Watson defeat a chess Grandmaster and then win on “Jeopardy.” However, now Watson needs to make the journey to operate in the real world of business problems.
These use cases and applications are still undefined and will emerge over time. It is, in fact, the challenge of problem definition and incremental adoption that stands in the way of progress. It’s easy to imagine that there are limitless applications for Watson; but for Watson to take off quickly, we need to identify big issues with large payoffs. Without these game-changing applications we will wait for several years for cogitative computing to make the contribution that it is clearly capable of.
To recover its billion-dollar investment and create a market for cognitive computing, IBM has every incentive to hasten the adoption. However, it has yet to identify the break-through problems that will drive rapid adoption. It is all very well to believe that the power of the technology will inevitably drive adoption; but if cognitive computing is like other disruptive technologies, it will come slowly and in spurts.
To hasten adoption, my best – and unsolicited – advice to IBM is to identify big business problems where Watson can make a structural change and drive massive benefits. Clearly, working as a companion to oncologists is such an area. And given that healthcare is 20 percent of the U.S. GDP that alone may be worth the journey.
But for enterprises beyond healthcare, I feel challenged as to what other big structural changes Watson and cognitive computing could provide.
I strongly suggest that you find a way to experience Watson’s power. It’s is so powerful that I, like IBM, am struggling with where we should take it.
As I’ve pondered its possibilities, I think underwriting and the claims process in the insurance sector holds tremendous opportunities. And within IT, I think the service desk and problem solving that IT departments contend with could be dramatically enhanced with this technology. With a cognitive computing tool as their companion, they could deliver a higher quality of service and greatly improve productivity. Clearly the area of security would benefit substantially as we find ways to keep the black hats out of our data.
I’m very interested in other points of view as to where we can put cognitive computing to work, so please add your comment below.
Photo credit: Wikipedia