A five-step formula is available for data and analytics leaders to quickly implement AI and realize value.
Transitioning from a desire to utilize AI to a practical strategy may seem daunting, but it need not be. Often, data and analytics leaders become sidetracked by selecting AI tools and techniques without defining their goals. With 32% of respondents in the 2023 Gartner CIO and Technology Executive Survey reporting that their organizations have deployed AI and ML, and an additional 17% planning to follow suit within a year, D&A leaders will need to develop a data-oriented AI strategy. To achieve quick time-to-value, this five-step formula provides a tactical approach for D&A leaders to begin with AI.
The first step involves identifying use cases that are meaningful, measurable, and can yield quick wins
Starting with the end goal rather than the initial question is a common approach in successful AI and ML initiatives. Every AI project should have a defined business impact and measurable outcome, which can evolve as the project progresses. The first step is to select AI use cases by identifying critical problems that traditional techniques have failed to solve. Consider both feasibility and business value when evaluating opportunities for success, while also keeping in mind the importance of time. To achieve quick wins, limit the project scope to a nine-week timeline, which has proven successful for many MVPs. D&A leaders should be able to articulate the value proposition of the selected AI use case and establish metrics that demonstrate its impact on business and financial performance. Beyond accounting metrics, the MVP's effect on business, information, and stakeholder value should also be considered.
The second step involves assembling a project team for the AI initiative.
The AI team should consist of three personas that balance each other out: an AI specialist with expertise in multiple AI techniques, such as ML, rule-based systems, or NLP; an IT professional with an understanding of current IT capabilities; and a subject-matter expert with knowledge of business requirements and metrics. Each persona requires a range of skilled professionals. Despite common misconceptions, AI skills are not necessarily rare, costly, or enigmatic. For instance, database administrators or mathematically inclined data engineers can become proficient data scientists. The ideal team members possess solid ML knowledge, a strong grasp of statistical principles, a keen interest in data exploration, and excellent collaboration skills. Motivated, open-minded, competent, and focused champions, typically found within the organization, make for a successful AI MVP.
The third step involves collecting the required data for the AI project
Another misconception surrounding AI is that large amounts of data are necessary for building AI models. However, many successful AI use cases require only a moderate amount of data, as long as the data is relevant and of good quality. In contrast, poor data quality is a sure path to failure for any MVP. The amount of data required may differ depending on the AI techniques used. For instance, ML techniques usually need more data than optimization or logic-based techniques. Nonetheless, the quality and relevance of data should always take precedence over quantity for the chosen use case.
The fourth step involves choosing AI techniques that are aligned with the use cases, the team's skills, and the available data
Choosing the right AI techniques for a specific problem depends on various factors, such as the available data and the skills of the team. For instance, ML techniques are suitable for uncovering patterns in large datasets but require high-quality data and expertise in analytical mechanisms. In contrast, optimization techniques may be more appropriate for managing gate and crew assignments during a snowstorm but require skills in operations research and data gathering. Additionally, other technical considerations, such as IT infrastructure resources and user interfaces, should also be discussed at this stage to ensure a successful MVP.
Transitioning from a desire to utilize AI to a practical strategy may seem daunting, but it need not be. Often, data and analytics leaders become sidetracked by selecting AI tools and techniques without defining their goals. With 32% of respondents in the 2023 Gartner CIO and Technology Executive Survey reporting that their organizations have deployed AI and ML, and an additional 17% planning to follow suit within a year, D&A leaders will need to develop a data-oriented AI strategy. To achieve quick time-to-value, this five-step formula provides a tactical approach for D&A leaders to begin with AI.
The first step involves identifying use cases that are meaningful, measurable, and can yield quick wins
Starting with the end goal rather than the initial question is a common approach in successful AI and ML initiatives. Every AI project should have a defined business impact and measurable outcome, which can evolve as the project progresses. The first step is to select AI use cases by identifying critical problems that traditional techniques have failed to solve. Consider both feasibility and business value when evaluating opportunities for success, while also keeping in mind the importance of time. To achieve quick wins, limit the project scope to a nine-week timeline, which has proven successful for many MVPs. D&A leaders should be able to articulate the value proposition of the selected AI use case and establish metrics that demonstrate its impact on business and financial performance. Beyond accounting metrics, the MVP's effect on business, information, and stakeholder value should also be considered.
The second step involves assembling a project team for the AI initiative.
The AI team should consist of three personas that balance each other out: an AI specialist with expertise in multiple AI techniques, such as ML, rule-based systems, or NLP; an IT professional with an understanding of current IT capabilities; and a subject-matter expert with knowledge of business requirements and metrics. Each persona requires a range of skilled professionals. Despite common misconceptions, AI skills are not necessarily rare, costly, or enigmatic. For instance, database administrators or mathematically inclined data engineers can become proficient data scientists. The ideal team members possess solid ML knowledge, a strong grasp of statistical principles, a keen interest in data exploration, and excellent collaboration skills. Motivated, open-minded, competent, and focused champions, typically found within the organization, make for a successful AI MVP.
The third step involves collecting the required data for the AI project
Another misconception surrounding AI is that large amounts of data are necessary for building AI models. However, many successful AI use cases require only a moderate amount of data, as long as the data is relevant and of good quality. In contrast, poor data quality is a sure path to failure for any MVP. The amount of data required may differ depending on the AI techniques used. For instance, ML techniques usually need more data than optimization or logic-based techniques. Nonetheless, the quality and relevance of data should always take precedence over quantity for the chosen use case.
The fourth step involves choosing AI techniques that are aligned with the use cases, the team's skills, and the available data
Choosing the right AI techniques for a specific problem depends on various factors, such as the available data and the skills of the team. For instance, ML techniques are suitable for uncovering patterns in large datasets but require high-quality data and expertise in analytical mechanisms. In contrast, optimization techniques may be more appropriate for managing gate and crew assignments during a snowstorm but require skills in operations research and data gathering. Additionally, other technical considerations, such as IT infrastructure resources and user interfaces, should also be discussed at this stage to ensure a successful MVP.
The fifth step involves Establishing the Framework for AI Proficiency
Implementing a few MVPs across different business problems can help organizations identify gaps in skills, data, technology, culture, readiness, and AI education. Once sufficient AI knowledge has been accumulated, consider structuring it into an "AI lab," "AI competency center," or "AI center of excellence." This approach enables AI experts to maintain close ties with the business, stay current on the organization's technical capabilities and investments, and remain aligned with the overall AI strategy. Experts can participate in "tours" within lines of business, working alongside subject-matter experts and IT professionals on various projects for months at a time.
Following these steps, the organization should aspire to a more ambitious AI vision. This approach will enable organizations to begin working with AI, identify strengths and weaknesses in technology and culture, and determine the desired business outcomes that will inform the longer-term strategy. After gaining experience, a formal AI strategy can be developed and transformed into an "AI first" strategy.
Erick Brethenoux is a Distinguished VP Analyst at Gartner, Inc., who specializes in machine learning, artificial intelligence, and applied cognitive computing. At the Gartner Data & Analytics Summit in Orlando, Fla., taking place from March 20-22, Gartner analysts will be offering additional insights on AI.
Following these steps, the organization should aspire to a more ambitious AI vision. This approach will enable organizations to begin working with AI, identify strengths and weaknesses in technology and culture, and determine the desired business outcomes that will inform the longer-term strategy. After gaining experience, a formal AI strategy can be developed and transformed into an "AI first" strategy.
Erick Brethenoux is a Distinguished VP Analyst at Gartner, Inc., who specializes in machine learning, artificial intelligence, and applied cognitive computing. At the Gartner Data & Analytics Summit in Orlando, Fla., taking place from March 20-22, Gartner analysts will be offering additional insights on AI.