Editor’s note: The following is a guest post by Annette Zimmermann, VP analyst at Gartner.
Enterprise executives deem generative AI a key focus area in 2024 and beyond, with spending on the technology projected to reach $443 billion by 2027. To prepare for the adoption journey ahead, CIOs are looking for AI expertise to prevent their companies from falling behind.
In most cases, this means looking towards an AI service provider for help.
AI service providers are helping in areas that include strategy, data engineering, model development and testing, operations and governance across a wide variety of AI technologies, including both symbolic and machine learning constructs.
Data readiness is a key term that AI service providers are using as they look to help their clients through a successful generative AI journey. From the perspective of the AI service provider, it is one of the main barriers to AI implementation. From the perspective of the CIO, hallucinations produced by generative AI models at the beginning of a project are among the biggest technical challenges.
Therefore, it is imperative for CIOs to explore data services and responsible AI as two foundational concepts, which is where an AI service provider comes in.
The nature and extent of data challenges at the client side cover a wide spectrum (for example, data silos, outdated content architecture and lack of data science skills). These issues must be addressed before embarking on a generative AI journey.
There are two factors that drive the need for CIOs to utilize data services from an AI service provider:
- Making the existing data actually usable for AI.
- Increasing data literacy and enabling citizen data scientists among adopter organizations. This allows employees across the organization to leverage the technology.
With focus especially on the latter, there is clear evidence that data centricity throughout the user organization builds the foundation for an AI-centric future business.
Addressing the data services gap
There is a significant opportunity for AI service providers to offer data services that will impact a client organization on several levels.
This is not only about unlocking the gold mine of data that every adopter organization possesses, but also about the democratization of data science and helping adopter organizations to achieve “data readiness.” Especially for larger enterprises, a comprehensive data service offering has become a point of differentiation.
Some of the larger AI service providers are therefore leading with data service capabilities, which have become an integral part of their value proposition. The data services they provide to set enterprises up for continued success can be categorized into four groups:
- Data modernization: data integration, cloud data migration and data management.
- Data architecture: infrastructure modernization, containerization and deployment, and security and access management.
- Data science: data science as a service, and advanced analytics and informatics.
- Data engineering: data sourcing, curation and labeling, and synthetic data generation.
Data quality, suitability, imbalance or mislabeling are the main causes of hallucinations. It is therefore crucial to utilize these data services to develop mechanisms and responsible AI frameworks to reinstall trust among client organizations.
With this in mind, CIOs should ensure they are accomplishing the following tasks to address these problems when working with an AI service provider:
- Focus on becoming a data-centric business by developing frameworks and best practices for their data science program that will enable generative AI at scale.
- Establish a list of use cases for generative AI and a method to evaluate new use cases. This should include methods to evaluate the data availability and quality for the intended AI use case.
- Look for transparency and proactivity regarding responsible AI practices by following a trust, risk and security framework to reinforce trust in generative AI technology.
With AI service providers ramping up generative AI projects, client organizations must address their challenges with data readiness. In this crowded field that also has significant technical challenges, enterprises must look to data services as part of AI services to ensure project success.
