Painting a picture of scalability with AI Canvas
As communication service providers increase their investment in artificial intelligence (AI), Dr. Tayeb Ben Meriem, Orange’s OAM Senior Standards Specialist, explains how TM Forum's AI Canvas gives communications service providers (CSPs) the practical help they need to introduce AI solutions at scale.
26 Aug 2021
Painting a picture of scalability with AI Canvas
As communication service providers (CSPs) increase their investment in artificial intelligence (AI), Dr. Tayeb Ben Meriem, Orange’s OAM Senior Standards Specialist, explains how TM Forum's AI Canvas gives communications service providers (CSPs) the practical help they need to introduce AI solutions at scale.
IDC estimates that the worldwide spend on AI software, hardware and services will grow 15.2% during 2021 to reach $341.8 billion by the end of the year.
CSPs are among the companies increasingly using AI to solve their business problems. As their use of AI grows, so will their need to streamline and standardize how they introduce and use AI software.
Orange is one of several telcos that have been heavily involved in developing TM Forum’s AI Canvas. Its aim is to make more standardized, scalable and efficient use of AI software to resolve business problems.
“We do not want AI to become a new silo,” says Ben Meriem. “Our main requirement is to have standardized solutions that we can integrate easily.”
Yet successful standardization depends on being able to take into account the wider operational and business environment.
“In order to take full advantage of AI in an efficient and safe manner in their business at large scale, CSPs have to consider the whole AI journey and lifecycle management – strategy, procurement, deployment and operations through DevOps /AIOps/TestOps – as a continuum and as a nested feedback, closed loop-driven process that should have minimal or no human intervention,” says Ben Meriem. However, “the most challenging aspects that are specific to AI are the availability of frameworks, tooling, means and assessment metrics.”
In short, CSPs need to be able to make AI explainable and demonstrate that it is being used fairly and ethically, and they need ways of managing its use at scale in order to do so.
The first step is to determine whether a particular business problem requires an AI-based solution, explains Ben Meriem. Once this is done, it is important to evaluate how the AI solution will perform in a production environment, which is where TM Forum’s AI Canvas comes in.
The AI Canvas draws on TM Forum’s widely used Business Process Framework (eTOM) to let CSPs map and understand AI processes at scale.
“We are already using TM Forum’s business canvas, so basing the AI Canvas on this makes our stakeholders comfortable and helps adoption,” says Ben Meriem.
The AI Canvas offers CSPs a reproducible, generalizable and scalable framework to accommodate multiple use cases, which can be mapped to areas identified in the Business Process Framework model. This allows users to center their problems around key themes and operational areas, and also to deconstruct functions for analysis.
“When we have the real input of use cases, it is easier to find commonalities and then create the abstraction layer for logical and operational architectures,” explains Ben Meriem.
Orange has already customized and operationalized TM Forum’s AI Toolkit for some internal projects and is assessing AI Canvas as part of TM Forum’s AI Toolkit to two existing AI use cases: anomaly detection and root cause analysis for FTTH (fiber to the home) and for RAN (radio access networks).
Find out more about TM Forum's AI Canvas.
IDC estimates that the worldwide spend on AI software, hardware and services will grow 15.2% during 2021 to reach $341.8 billion by the end of the year.
CSPs are among the companies increasingly using AI to solve their business problems. As their use of AI grows, so will their need to streamline and standardize how they introduce and use AI software.
No more silos
Orange is one of several telcos that have been heavily involved in developing TM Forum’s AI Canvas. Its aim is to make more standardized, scalable and efficient use of AI software to resolve business problems.
“We do not want AI to become a new silo,” says Ben Meriem. “Our main requirement is to have standardized solutions that we can integrate easily.”
Yet successful standardization depends on being able to take into account the wider operational and business environment.
“In order to take full advantage of AI in an efficient and safe manner in their business at large scale, CSPs have to consider the whole AI journey and lifecycle management – strategy, procurement, deployment and operations through DevOps /AIOps/TestOps – as a continuum and as a nested feedback, closed loop-driven process that should have minimal or no human intervention,” says Ben Meriem. However, “the most challenging aspects that are specific to AI are the availability of frameworks, tooling, means and assessment metrics.”
In short, CSPs need to be able to make AI explainable and demonstrate that it is being used fairly and ethically, and they need ways of managing its use at scale in order to do so.
Scaling AI
The first step is to determine whether a particular business problem requires an AI-based solution, explains Ben Meriem. Once this is done, it is important to evaluate how the AI solution will perform in a production environment, which is where TM Forum’s AI Canvas comes in.
The AI Canvas draws on TM Forum’s widely used Business Process Framework (eTOM) to let CSPs map and understand AI processes at scale.
“We are already using TM Forum’s business canvas, so basing the AI Canvas on this makes our stakeholders comfortable and helps adoption,” says Ben Meriem.
The AI Canvas offers CSPs a reproducible, generalizable and scalable framework to accommodate multiple use cases, which can be mapped to areas identified in the Business Process Framework model. This allows users to center their problems around key themes and operational areas, and also to deconstruct functions for analysis.
“When we have the real input of use cases, it is easier to find commonalities and then create the abstraction layer for logical and operational architectures,” explains Ben Meriem.
Orange has already customized and operationalized TM Forum’s AI Toolkit for some internal projects and is assessing AI Canvas as part of TM Forum’s AI Toolkit to two existing AI use cases: anomaly detection and root cause analysis for FTTH (fiber to the home) and for RAN (radio access networks).
Find out more about TM Forum's AI Canvas.