With the growing demand for data science and cognitive technology approaches across industries, organizations are learning how to successfully implement and manage newer and smarter tools and systems. What are the challenges companies face when adopting AI and ML models for their organizations, and how can teams work to overcome these obstacles?
A tan next Data for AI eventIn particular, Anil Kumar, Executive Director – Head of AI Industrialization at Verizon, will share the ways Verizon has leveraged AI to overcome some of its key challenges. Last January, the Machine Learning Lifecycle 2021 conference brought together Radha Sankaran, Executive Director of Algorithmic Customer Experiences at Verizon Wireless, where she shared an overview of the current state of AI use and its challenges, techniques and impacts. At the next Data for AI virtual event, Anil Kumar, also from Verizon Wireless, will talk more about his experiences.
The state of AI adoption and its challenges
Two of the main parts of Verizon include its more traditional, home or wired side of the business, as well as its wireless side. The current state of AI adoption within the enterprise can be described as being in a cycle of medium maturity, with the wired side having started earlier in more advanced digitization due to the need to be innovative in cost control. On the wireless side, a significant transformation took place from 2019 and 2020 with the creation of separate organizations of Chief Data Officer and Chief Customer Officer. The CDO organization is responsible for all data-related projects, and this is where all data scientists reside. The client organization then focuses on adopting the best practices developed in the home side of the business and extending them to the wireless side through constant analysis. The organization also feeds all the requirements into the organization with data to run and improve. Working together, these organizations are advancing and overseeing the growing adoption of AI in the mobile space for Verizon Wireless.
Businesses can face many challenges when transitioning to AI and ML-based technologies, especially when there are already well-established traditions within large-scale operations and teams. One of the biggest organizational challenges faced by the Verizon team was to tie together all the existing data related to the customer experience. Since the organization had previously looked at all experiences in a channel-driven fashion and optimized the channels separately, all data was also siled. So figuring out how to pull all the isolated data together into a cohesive omnichannel that allows the team to approach customer experiences in a truly holistic way posed the biggest challenge. Additionally, in terms of corporate culture, the establishment of a CDO organization and CXO organization moved the owners from outside from separate channels to a centralized organization which enabled the aforementioned holistic approach. For the Verizon teams, it was about breaking down silos and bringing together the different organizations like the CDO, CXO, and global technology departments to discuss issues and solutions. Together, the teams are focused on putting in place the right data architecture, governance, and products in the most efficient way.
Development and management of AI models
From a more technical perspective, building and implementing AI models requires the right people and the right tools, which are either developed within organizations or outsourced. For Verizon, the acquisition of Yahoo in recent years has provided an excellent basis for accelerated deployment of AI. Since Yahoo organizations already had very mature data lakes and the respective ML tools, teams at Verizon were able to use the pre-existing CI and CD pipeline setups to build models from the lower layer and perform a real-time scoring. Due to this smooth transition, no outsourcing was necessary; everything was available internally.
Unlike more traditional software which can essentially be left on its own after initial implementation, AI models require ongoing monitoring and maintenance to prevent model degradation or drift. At Verizon, organizations call personalized customer experiences propositions, and these data-driven and AI-driven events need to be constantly monitored for optimization and performance evaluation. Using both adaptive and predictive modeling, it’s important for teams to identify whether the issues that arise are related to the models themselves or due to data gaps or data quality issues. Using both flow data and batch data, Verizon organizations are able to train their AI models continuously through real-time data feedback loops.
Impacts of the pandemic and future obstacles to overcome
While the recent pandemic caused a great deal of disruption in nearly every industry around the world, for Verizon it resulted in a significant benefit: the pandemic prompted Verizon to become a digital-centric company. Before the pandemic, much of the sales would be in retail stores, but the company had to find a way to transform itself into a digitally-driven organization almost overnight. Digital adoption in less traditional areas such as retail and telesales organizations had to be accelerated along with the rest of operations, and teams at all levels had to come together to cope with this transition. With this change, organizations like the CDO and CXO teams are able to advance digital adoption and further leverage data and AI.
Looking to the future in a post-pandemic world, perhaps the biggest barrier to large-scale AI adoption lies in the data itself. Securing access to the right data, generating the best methods for analyzing that data, and ultimately understanding how all the necessary components and information come together remain key challenges and goals for Verizon, and many other businesses on a scale. similar, something Anil will discuss at Data for the AI event of December 2, 2021.