As a business leader, you know that AI adoption will be key to achieving your future growth objectives. But exactly how to use AI to drive value is probably less clear. A recent article in The Economist revealed that most American companies are still in the experimental phase, with only about 5% actually using AI to produce goods and services. The evidence suggests the leaders are reaping the rewards: according to McKinsey, companies with mature AI capabilities are outperforming laggards by two to six times.
The challenge today, even for the most advanced users of AI, is to scale beyond islands of excellence to achieve broad impact on business metrics. To make this leap, three important constituents need to work together: business leaders; CIOs and the IT function; and data science and analytics experts. Each has a critical role to play, but none can operate in a silo. Communication, cooperation, and an understanding of the needs of the others will all be fundamental to successfully delivering AI at scale.
Move past your assumptions
Business leaders tend to make certain assumptions about AI that can undermine efforts to transition from experimentation to at-scale adoption. This ultimately leads to finger-pointing and blame when AI doesn’t deliver. First, it’s tempting to view AI as a magic bullet or miraculous medicine that can cure all ills. Whatever the business problem, just “getting some AI” will solve it. Also, some think of AI as a technology that can be bolted onto existing systems and processes—a perception that underestimates the work required of business teams.
To capitalize on AI, business teams need to show leadership and proactively take responsibility for AI deployment in their departments. This must be a business solution-led approach, not a technology-led implementation. Only those at the coalface can precisely define how and where AI can help meet business goals. Setting clear objectives for AI implementation will enable an effective AI strategy and avoid the development of ad hoc AI tools that cannot scale to meet real-world business requirements.
Treat AI like a puppy, not a toy
Standalone AI tools may look shiny and attractive, but there are still only a few applications where AI is plug-and-play. To ensure AI is adopted successfully, business leaders must consider how it will be used and integrated into existing projects and processes. Will AI create outputs for direct, unsupervised decision-making, or will it feed into and combine with other inputs to inform internal processes? Are AI models a solution or part of the solution?
Finally, business teams need to understand that AI is an ongoing commitment. Think of it more like adopting a puppy than a buying a toy. Aside from prepackaged generative AI-powered tools like Copilot, AI is not something that business users can just acquire and then pick up and use when needed. AI models by nature require constant attention. They need to be fed vast amounts of quality data and continuously monitored. Who will be watching to ensure outputs are within expected parameters? Whose job is it to spot and report eccentric behavior and hallucinations? It’s business users who have the knowledge and experience to spot these aberrations, so they must be properly resourced and equipped to take responsibility for ongoing management.
Foundations for successful AI deployment at scale
Implemeting AI across the enterprise requires “rewiring” the organization, so business leaders will need to work closely with IT and analytics teams to ensure that data, IT resources, and business demands are continuously aligned. And AI requires a solid data foundation, so it’s critical for business teams to develop a data-centric mindset and understand their role in creating and curating data as an enterprise asset. They must emphasize the value of high-quality data and leverage a unified data architecture to enable data harmonization and efficient data use and reuse.
Business teams must also be aware that AI models drift and may require retraining. Developing this understanding requires some knowledge transfer between analytics and business teams, which will help both sides recognize the cost and value of requests and commitments. After all, AI cannot operate in a vacuum, and the same cost controls should be exercised as for any other expense.
As an emerging technology, AI is still in an early innovation phase. As companies explore its potential, it’s to be expected that not every experiment will pan out as hoped. But by taking a proactive role, establishing trusted lines of communication, and fostering cooperation instead of playing the blame game, business leaders can ensure setbacks are leveraged as opportunities to learn and improve, paving the way to achieving the promise of AI.
