Amdocs : AI presents a new opportunity to invest in CSP data

An organization's readiness for adopting artificial intelligence (AI) capabilities is only as good as the weakest link in the chain. Accepting this view means that developing capability in each of the complementary business and technology areas necessary to enable AI programs is a more-than-the-sum-of-its-parts calculation. Each pillar of strategy, organization, technology, data, and operations is interdependent, and progress must be made across the board to lay the right foundations for building and benefiting from AI solutions. While data has enjoyed time in the executive limelight, for many it remains stubbornly challenging to source, manage and access - the rise of AI presents a golden opportunity to make new investments in data management technologies, while helping realize the business' AI aspirations. AI use cases need data to be successful Data-driven decision making is a business good, and while gut instinct still has a role to play - the majority of critical decisions demand evidence derived from historical data, matched to predictions made from it. AI is an extreme use case for data, AI doesn't have gut instinct, and is only as good as the data it uses to achieve its goal. Whether it's a chatbot providing simple customer service, or a machine learning algorithm looking for maintenance events across a landscape of Internet-of-Things connected machinery, the availability and quality of data will define the level of success it achieves. The enabling data capabilities focus either on the people, process, or technologies employed by an organization to manage its data. Typically, Ovum's experience is that while technologies to manage data are - to varying degrees of sophistication - in place, the processes that support them are often insufficient or broken, and the people-dependent aspects, such as senior executive sponsorship, an afterthought. An AI assessment should provide an understanding of some key data capabilities; helping identify those areas which require focus, and helping create practical steps for new investment of resource and effort to grow the availability, and reliability of data within the organization. While some immediate opportunities to make improvements may exist, for example the appointment of a senior executive with responsibility for data, likely a Chief Data Officer (CDO), for most the journey to improving data is a never-ending one. That doesn't make for comfortable reading, as with the prospect of any major program, it can feel overwhelming: where to start? Ovum strongly advises a dual approach to make this journey's start a little easier:

1. Identify the quick wins and demonstrate early proof of value - where the current state assessment surfaces areas for improvement, these should be carefully investigated and prioritized based on which efforts will bring the quickest and most visible return on investment. Adopting this approach makes for easier investment cases, and a 'snowball' effect of continuous improvement delivered by investment in data - building trust, and making future, more complex requests easier to justify.
2. Partner with the AI program to understand and meet their needs - the benefit of proximity to high-value strategic programs like AI is generally an easier investment environment. The approach in point one should not, however, be sacrificed, it will be just as critical to demonstrate rapid return on investments. Working with these programs also brings further clarity to which areas require the most immediate focus; rather than attempting to change everything at once, the data most important to that program can be identified and tackled first.

It is important to remember AI is not the only game in town, and data management is required for multiple use cases across and organization, not least financial management and compliance / regulatory reporting. Balancing the needs of these data mission critical requirements with new, value-adding use cases like AI will be challenging: the role that a CDO plays in this regard cannot be underestimated. A final thought must go to what happens when data is either insufficient or of poor quality when used in AI use cases. The best-case scenario is one of failure to return value from the use, for example, where a network optimization machine learning algorithm is fed incomplete data and cannot perform effectively. The worst-case scenario should be more worrying, there are already multiple examples of AI chatbots running amok on social media to consider. Senior executives who are reticent about signing checks for the necessary data work to enable AI programs should be aware of the very real, and high-profile risks of not getting the data right. About the author: Tom Pringle, Head of Applications Research, Ovum. Tom focuses on data and information management, analytical technologies, and the evolution of enterprise applications. His topic expertise lies in digital transformation, automation, and customer data. For Ovum's enterprise clients, Tom provides guidance on the effective use of data and analytical technologies and the role they play in broader transformational programs. This research encompasses the theory and practice of data architecture and governance, the analytic tools used to explore data, and the importance of data privacy and ethics.