Monday, March 20, 2017
I was at a conference where an animated, heated, passionate discussion was happening between members representing the technology players’ leadership team and some customers. It was quite obvious that there was a difference of opinion leading to a disagreement on the subject. There were multiple points of view which added to the liveliness of the debate. The conference organized by an independent organization was on one of the currently hot technologies – artificial intelligence and its manifestation in the form of bots.
The Structure of words determine segmentation methodology and parsing for semantic analysis, tasks that can now be done with higher accuracy though complete sentences require additional facts and external world knowledge. For written text in a chat session, the bot is able to hold fort quite successfully when addressing well-defined tasks and decision trees. Many online portals and businesses have already deployed chat bots to supplement agents who step in when the bot is unable to parse and respond to a question.
Eliza was seen as a breakthrough, so are personal assistants who respond to and act upon simple tasks using basic language parsing tools. While the technology is nascent, attempts to make computers chat with humans have been around for over half a century. Current experiments indicate that they can be purposed for specific tasks. Excitement revolves around automated interactions to improve efficiency and reduce cost in comparison to current models of humans talking to customers for customer support issues.
The Natural Language Processing (NLP) journey that began with SHRDLU has improved significantly with Machine Learning and Deep Learning. The ability to pass the Turing Test (the test investigates whether people can detect if they are talking to machines or humans) is still some time away, though we are getting closer to the milestone. Though some may claim that Eugene passed the test a couple of years back, many do not agree with the results. Artificial Intelligence barrier is yet to be crossed convincingly.
As bots mature and they are evolving fast, their ability to manage specific tasks is already giving organizations the benefit of consistent responses to basic and mundane queries; use cases around query and response of HR systems on leave balances, tax options, and others have been successfully deployed. Externally airlines and hospitality industry have taken the lead while inside our homes personal assistants are making an appearance; the biggest driver however has been the smartphone listening, responding and taking actions.
Call centers and IVR systems continue to drive customers crazy with long decision trees and stupid obvious questions; wait times add to the woes of the frustrated customer. Offshore call centers did reduce the cost but faced backlash on other social impact it created. Bots promise to take up the challenge and address the customer with agility and surprisingly improved outcomes. The rich repository of information and past interactions helps the rules engine train effectively and take decisions objectively.
As the cost of deployment continues to fall with mass development and niche players, companies are slowly embracing this revolution. Which brings to fore the point that will they be able to replace humans in call centers as the bot learns and graduates to the next level of interactions ? Recent times have seen the rise of hullabaloo about jobless growth and the disappearance of low end jobs to be replaced by machines; as the gap to the Turing Test reduces, the probability that they can starts staring us in the face.
The job loss is indeed going to impact destinations that build their business models around calls and low end business process outsourcing (BPO). Many existing players have already started upgrading skills as well as retrain staff for other roles; enterprises will ruthlessly choose efficiency over distant loss of employment even if it involves initial high investment. Are there options available to current players ? It is a race against time for incumbents to move up the value chain or become irrelevant to their customers.
Unless, unless they embellish their services with bots and attack their own business before someone else does. A handful of providers with strong technology backing from within or their parent companies are beginning to offer value added services using bots thereby changing their customer acquisition and retention strategies aligned to the new reality. They are using their rich knowledge repository to train the technology solution towards addressing existing and new problems thus opening up opportunities.
A system is as good as the people who build it; in the end it will be humans who will continue to create the differentiator !
Monday, March 13, 2017
Rummaging through my archives I came across a presentation I had made in a large IT conference fifteen years back to the date. The subject line had me wondering if I had made a fool of myself in the gathering considering that the topic was not my core expertise, though I was a bit enamored by the discipline. Memory is kind and there is no recollection of being booed off stage or being in an uncomfortable position. The presumptuous title of the presentation was “How to protect your enterprise from being hacked”!
Organizations get hacked for many reasons, though most of the hacks in recent times were attributable to human error, lapse in controls, malice towards existing or ex-coworkers or bosses and finally social engineering resulting in compromised data which allowed nefarious elements to gain access and control of information assets for potential future misuse. There were also few brute force attacks as well as skilled hackers who could break the firewall and other technologies that protect the digital ecosystem.
The past decade and half has seen exponential growth in devices connecting to the internet; what started as basic email on mobile, extranets and the surge with the dotcom bubble has grown beyond the predictions of all kind of futurists and consultants surviving the blips due to dot bust and many years later the subprime crisis. M2M, IoT and connected consumer devices have already added to the exposed digital fabric available, vulnerable to attacks as well as errors and omissions by people who configure and monitor.
Back then before the turn of the century reported security incidents were a handful; current reality is 10X of that and for clarity these are reported numbers. Guestimates on the actual number portray a similar multiplier on the reported number. The difference lies in BYOD which has removed the mobile end user compute from purview of the enterprise subduing the number. Smartphones and Tablets, wireless hotspots, public internet kiosks and free terminals at airports, all have helped in accessing information anytime, anywhere.
For IT organizations threat vectors multiplied sending them on a quest for better security and balancing the demand and need for access to corporate systems. Controls and checks soon became bureaucratic with everyone wanting to connect as a result of undue corporate pressures. The number of breaches continues to rise with IT security playing catchup. MDM anyone ? Locked USB ports, containerized phones, IRM enabled documents, the world has changed while we continue to stay exposed with cookies/mobile app trackers.
Globally, Governments have giving a thrust to digital e-governance and citizen services; identities and records of interactions with Government, tax filing, health records, bank statements, what have you, almost everything is digitized across most countries with varied degrees of information security policies, processes and technology. Access via mobiles and apps is the base expectation which needs to be fulfilled; feature phones too have been enabled using USSD (Unstructured Supplementary Service Data) which can be compromised.
Interestingly time to uncover a breach has now increased from weeks to many months and in rare cases more than a year; this rise is despite availability of plethora of solutions. Software is getting bulkier, crammed with features; integration with other solutions is now the norm exposing solutions with potentially unfixed or insecure APIs (application programming interface) from third parties. Unfortunately security wrappers and multi-factor authentication make solutions unwieldy or complex to end users.
Fifteen years back the discussion was about security policies, management endorsement and budget allocation; it was about protection from insiders – disgruntled employees and contractors. One of the key elements of an information security strategy was education of involved stakeholders, their responsibilities, dos and don’ts. Today is still about security policies, more controlled that clearly separate the personal from enterprise; earlier digital access was controlled by hierarchy and exception, today exceptions to the rule for business is the norm.
I wish I could give a presentation today with the same confidence and aplomb that I did decade and a half back; technology has swamped our lives with blurring boundaries between technology at work and personal use. The continuum with high dependence on devices and tether to the internet for almost everything, logs our daily activities in the background only to be used against us. Enterprises struggle to create a balance between storing data on the cloud and enterprise vaults only to discover that neither are safe.
Reality is that most of your data is out there available for pennies to whosoever wants it whether you like it or not !