Internet of Everything

  • 1.  AI Improving Customer experience

    Posted Sep 19, 2017 15:03
    How are the Telecom service providers gearing up to harness the power / insights generated from AI to improve customer experience and drive towards disproportionately leaner & agile organisations.

    ------------------------------
    Mukul Gupta

    ------------------------------


  • 2.  RE: AI Improving Customer experience

    Posted Sep 20, 2017 01:08
    Yesterday Ooredo Maldives launched a customer service app that uses facebook messenger. Many operators are moving to digital tools particularly conversational tools that can accommodate automation and AI support over time. I can see this enabling product provisioning as well as reducing support overhead but the biggest benefit is engagement with customers at the opportune time with minimum fumbles.

    AI to me is even more useful internal to Telcos. From preempting network events and fault tolerance to identifying fraud, AI can be used in the network to improve up time and experience. Not much progress is being made at the moment but there is a clear path towards this in the future.

    Customer facing OSS/BSS operations will change sooner as CRM, ordering and services move to digital and take AI support to reduce overhead, and provide real time support to staff. Marketing will benefit from big data and analytics support including predictive scores and next best actions automated in conversations.

    2018 will be the year of a lot of partnerships and experimentation with digital technology for telecoms.

    ------------------------------
    Ehtisham Rao
    CEO, Servup
    ------------------------------



  • 3.  RE: AI Improving Customer experience

    Posted Sep 20, 2017 03:42
    Today the BBC News includes a pre-announcement of a speech by Theresa May telling tech companies to use AI to control what they mediate.

    Well, coming from her I suppose it is a bit like her waving a limp washing-up rag at a dirty dish.

    But it does raise the prospect that AI-based censorship facilities may become a side-effect development of attempts like this to curb misuse of the internet for propagandist purposes. Where will it end? I suppose some censorship could, in some circumstances, improve customer experience. (For example by the removal of unpleasant language or offensive content.) But there is a thin line between control meant to enhance customer experience and what may become a tool of government manipulation and social engineering.

    There is no doubt that Machine Learning offers a great deal of potential to control what gets through and what doesn't. But how can it know what is good customer experience and what isn't? Already we have media tycoons vying to have such control over public opinion. Giving them clever machines could make things worse.

    Whatever we choose to mean by "AI," I propose it should be to put power into the hands of the end-user customers - Not to deprive them of ANY experience.

    Perhaps there should be legislation to ensure that, wherever AI is used, its ultimate control lies in democratic choice, subject to general open mores. This would make it incumbent upon telco providers to open up all AI-based services they put into operation to public scrutiny and control.

    ------------------------------
    Peter Duxbury-Smith
    Intelligent Solutions Ltd

    ------------------------------



  • 4.  RE: AI Improving Customer experience

    Posted Sep 20, 2017 10:41
    One of my client is actively engaged in applying AI in assessing service And network-level incidents and events for predicting capacity bottlenecks and proactively avoiding outages - directly improving customer experience. 
    This is just one example...
    AI for predicting Early Life Failures, reducing Repeat Faults, optimising service operations are other use cases that will be very pertinent.

    ------------------------------
    Mandeep S Kwatra
    ------------------------------



  • 5.  RE: AI Improving Customer experience

    Posted Sep 21, 2017 03:40
    I used AI for an Event Management System I built for Vodafone 20 years ago. 

    It incorporated a Truth Maintenance System for hypothetical reasoning  to discover the actual status of network elements, given that there were 14 different databases holding inconsistent information about these elements. It also applied a form of Model Based Diagnosis to enrich the knowledge elicited from operators and represented in rules.  It used an event-driven Rule-Based Production System with added AI functionality I developed for the application. In the project I explored a variety of different forms of information correlation applied to several types of knowledge. Finding that topological knowledge of the managed network was the most useful. So that became the basis for guiding analysis by the system. There were plans to progress, in a later phase, to use some of the Machine Learning work I had done in a previous role developing projective industrial process control  (based on Dave Waltz's Memory Based Reasoning.) But that never happened because of organisational changes. (However, that Memory-Based technique pioneered much of the practical applications for Machine Learning we see in practice today.)

    There are many different approaches and techniques in "AI."  It is not a single solution. Not in the way so many seem to think it is, when they refer to using AI without saying which technologies from AI they mean. 

    So why the excitement about "AI?"  Why not try to establish what AI technology is actually applicable where?

    There is - and has been for a few decades - a real case for using some types of AI technique for some shapes of problem. This goes beyond using "AI" to jazz up interest for a project - Instead it requires analysis and understanding of the shapes of problems that require AI.

    So what are the shapes of telco problems that require AI? (Hint: They will be ones that are not soluable another way.)


    ------------------------------
    Peter Duxbury-Smith
    Intelligent Solutions Ltd

    ------------------------------



  • 6.  RE: AI Improving Customer experience

    Posted Dec 06, 2017 01:26
    This is great to know. Can you give some more details, like what kind of AI / ML algorithm they are using, what are the typical inputs and output etc

    ------------------------------
    Anirban Konar
    Cognizant Technology Solutions
    ------------------------------