Artificial intelligence has the potential to revolutionize financial services. According to the MIT Sloan and Boston Consulting Group’s 2019 Global Executive Study and Research Report, 90% of respondents agree that AI represents a business opportunity for their companies. Firms are investing heavily in AI capabilities, but few have a clear vision for their adoption strategy or a process for prioritizing projects, running experiments, and implementing AI on an enterprise-wide basis.
A recent survey conducted at the SIFMA Operations conferences earlier this year of over 200 representatives from banks, asset managers, broker-dealers, consultants, and regulators indicated that 78% do not have AI initiatives in production, and 20% are unsure or have no plans to do something in AI.
As effective use of AI becomes a necessity for financial firms simply to survive and stay relevant, taking an unstructured, wait-and-see approach is no longer an option.
Many firms implement point solutions without taking a strategic approach across the firm. Viewing AI adoption as an enterprise-wide transformation project in tandem with an overhaul of the firm’s digital and data capabilities can significantly increase the chances of a strong return on AI investments, while avoiding some of the hype around AI.
A common mistake companies make is placing too much focus on the technological aspects of AI; it’s important to identify the people, cultural, and change management factors that can help to ensure successful adoption of AI.
To help assist senior leaders with creating an AI strategy, five key components provide a blueprint for successful AI adoption: strategy, structure, systems, skills, and staff.
Taking an enterprise-wide view of AI, with top-down support from senior executives, sets the direction for how the firm will adapt to disruptive technologies. A clear AI strategy takes both a long-term and short-term perspective on AI — for example, considering how the technology could change business models, while also focusing on near-term opportunities to improve both internal efficiency and the customer experience with AI-driven products and services.
Many businesses fall into the trap of treating AI as a solution in search of a problem. This creates a vicious cycle where the initial use cases produce little value and dampen momentum for future investment. Businesses should identify their top three to five problems worth solving with AI that can drive the greatest value quickly, followed by a clear path to achieve on-the-ground impact. For example, if the effect of AI is saving a person’s time, reallocating that time by reengineering the process or the way teams are organized drives tangible value. This in turn encourages buy-in for future investment. Organizations must also strive to build AI solutions in a scalable and reusable manner so that they can be used throughout the firm rather than in specific business units or functions.
In addition to internal activities, AI strategy must also focus on developing ecosystems of external partners. For example, firms with scale across multiple clients and large datasets can offer innovative AI solutions while mutualizing the costs of adoption, providing a more cost-effective onramp to AI capabilities than building internally. According to the SIFMA Ops survey, more than 95% of respondents see value in co-developing AI with other firms and partners.
Finally, cybersecurity, information security, and data privacy are key guardrails of an AI strategy. A continual assessment of evolving regulation around AI is critical, especially ethical and explainable AI where end consumers are affected.
One of the common pitfalls to avoid in structuring an AI program is setting up an AI center of excellence and hoping it drives innovation. In its report, Driving Impact at Scale from Automation and AI, McKinsey observes that “organizations struggling to create value through analytics tend to develop capabilities in isolation, either centralized and far removed from the business or in sporadic pockets of poorly coordinated silos.”
Building cross-functional teams with those who understand the business, clients, and the industry, as well as those who understand AI technology, is much more likely to lead to genuine innovation. Once such teams demonstrate tangible value on initial use cases, they create a successful model that can be scaled across the organization.
Many firms try to develop AI solutions without having clean, centralized data pools or a strategy for managing them. Without this critical building block for training AI solutions, the reliability, validity, and business value of any AI solution is likely to be limited. Poor-quality data significantly raises the possibility of overfitting, model risk, and bias in results.
However, this does not mean that a business has to cleanse all available data before undertaking any AI initiatives, nor does it mean that a company needs to set up one unified data lake. McKinsey estimates that companies may be squandering as much as 70% of their data-cleansing efforts. The key is prioritizing these efforts based on what’s most critical to implement the most valuable use cases.
Setting up a technical architecture that can act as a sandbox to pilot and calibrate AI solutions before you launch it into production is an essential component in a firm’s AI strategy. Combining this with a reusable methodology for running experiments on data, assessing the performance of AI solutions against clearly aligned metrics, and then governing them on an ongoing basis is much more likely to yield successful results.
There has been a great deal of discussion recently about the democratization of data science. The concept includes giving a much wider range of employees at a firm basic training in areas such as data science and how AI works. Although not practical for every organization, it does make sense for firms to change the types of skillsets they cultivate in their staff.
Having some basic data science knowledge helps people who understand business processes and client needs come up with novel AI-driven solutions that can add value for the organization and its customers. Bridging the gap between technical experts and those with business domain knowledge through AI “translators” ensures that AI solutions spring from the convergence between imagination and understanding, rather than technical expertise alone.
Lack of focus on the human, cultural, and political dimensions of implementing AI is one of the key reasons projects fail to deliver the expected results. When company leaders define a change management program to communicate the benefits of AI to staff, they need to involve the employees who are most impacted and identify where they need to overcome cultural inertia to increase the likelihood of success.
Reimagining job design will also become more important as AI increasingly augments human performance. Helping staff to shift focus to higher-value tasks while also leveraging their knowledge of the business, clients, processes, and operations to continually govern and improve AI solutions will help workforces adapt to the widespread changes that AI will create.
Finally, having a defined plan to attract and nurture AI talent will allow an organization to retain people with the skill, intelligence, imagination, and understanding to create AI solutions that give it a competitive advantage.
Neha Singh is vice president of innovation and growth at Broadridge.