Editor's note: The following is a guest article from Moutusi Sau, a research VP on the CIO Financial Services Research and Advisory team at Gartner, covering banking, artificial intelligence trends and fintech.
Organizational interest in artificial intelligence (AI) projects has steadily grown in recent years. Just 14% of organizations had deployed AI in 2019; however, that number rose to 19% in 2020 and is expected to reach 24% in 2021, according to Gartner research.
This growing interest in AI will influence organizations' technology investments in the immediate future. Yet, AI adoption requires more than just the latest technologies or modeling techniques. When moving from a pilot AI solution into production — or scaling AI in the enterprise — CIOs need to articulate a clear business purpose and rationale to invest in these technologies.
There is no such thing as a one-size-fits-all business case for AI. Instead, the business case will be for a particular scenario, problem statement or use case that employs AI methods and techniques as part of the solution.
Here are the six factors that CIOs must consider when forming a convincing business case for AI, taking into consideration the differentiated obstacles organizations face on their journeys to AI adoption:
Factor 1: AI can seem costly without providing immediate ROI
Analyzing expected costs and benefits of a project is a critical component for any business case. However, in the case of an AI project, there might not be a straightforward answer. AI projects may appear costly without any immediate gains — particularly in organizations that are not used to setting aside budget for new business scenarios.
The return value of AI is closely intertwined with the aspirational value that the organization is seeking. Therefore, the ROI of AI projects are influenced by three key factors:
Strength in digital adoption: Organizations that are ahead in digital adoption have the most to gain from AI, as they have a natural advantage in digital transformation.
Seriousness in AI investment: AI investments cannot be perfunctory. Organizations that are benefiting from AI are investing seriously earlier than competitors.
Strong management support: This is closely interrelated with the culture of an organization. Most successful AI projects have always had management support for them.
CIOs must consider these factors when calculating the expected costs and benefits of an AI project. Be upfront with stakeholders that these costs might change considerably as the solution scope is explored and refined. Communicate a willingness to shut down AI projects where no clear benefit is emerging.
Factor 2: AI requires unique skills and talent
Talent acquisition is one of the top constraints that organizations face in AI deployment. Fulfilling talent needs is the most challenging for early-stage AI adopters, as organizations that have already achieved success are likely to combine hiring strategies that include internal and external AI talent.
CIOs should ensure AI talent is hired and developed from the very beginning of a project. To do so, devise a plan for acquiring and developing organizational AI skills that can be presented as part of the preliminary business case.
Factor 3: AI business cases need measurable value
Measuring the business value of AI projects early on is critical. In a Gartner survey, 39% of respondents that successfully deployed AI projects ran financial analysis on risk factors or conducted ROI analysis. For organizations to show that AI projects are much better than conventional technology methods, measurements become critical for buy-in and approval.
To determine the success of an AI project, there might be a need for multiple metrics that go beyond simple financial measurements. For instance, if an organization is using AI to increase the total customer count, interaction volume or customer interaction outcomes could be an additional measure of success.
CIOs should prioritize the value of measuring success from the start of an AI project. Be proactive in gathering data and consider including a mix of metrics that go beyond financial numbers.
Factor 4: The importance of data, training and algorithms
AI uses analytical algorithms to make sense of and act upon complex data. The interactions between data and algorithms are an essential component of an AI business plan. The effort of understanding, preparing and perfecting data for AI works beyond a single project and has a lasting effect by being usable for building many models.
Successful AI implementations include robust data and analytics infrastructure. CIOs must ensure business problems have enough supporting data for predictions, which contain patterns that executives expect to see in the future. For example, if the organization is making a prediction that could vary quarterly, data should span multiple years to be able to represent quarterly changes.
Factor 5: The decision to build, buy or outsource
The decision to build, buy or outsource is critically dependent on the resources available to the organization. The choice between these three options depends on the complexity of the projects, as well as factors like the IT department's level of maturity, the required time to solution, the urgency of the need and the organization's budget.
The best path forward is determined by the business problem the organization is trying to solve. To decide the right way forward, CIOs should use the following criteria:
Build if the proposed project will be unique to the organization and strong in-house data science skills are already available.
Buy niche applications that can be easily customized to suit organizational needs.
Outsource if the organization lacks both previous options but needs to start working on a project right away.
Factor 6: AI algorithms carry unique ethics and governance demands
AI is designed not only to support human decision-making; algorithms can also act autonomously. These AI algorithms, therefore, need to be trusted to act on behalf of all participant parties in each digital interaction.
The impact of ethical discussions in AI implementations is paramount since many AI-enabled systems rely primarily on machine learning based on underlying data. Without thinking about the ethical ramifications, AI systems can propagate undesirable behavior and affect the brand value of the organization. In this sense, business cases involving AI are often difficult to articulate, because AI systems can generate unpredictable or unexpected results.
CIOs should work with relevant stakeholders to start planning for ethics and governance demands when working on an AI implementation. Acknowledge the importance of ethics in AI implementations and be proactive in managing this challenge to build trust over time.