Trust is the missing—and essential link—in putting gen AI to work

While business leaders, managers, and frontline workers have a sense that gen AI can offer new ways to think, work, and create, they are still having difficulty figuring out how. Thus the gen AI paradox: according to a McKinsey survey earlier this year, 71% of companies are using gen AI in at least one business function, but few are seeing organization-wide, bottom-line impact. 

The fact is, putting new technology into people’s hands doesn’t mean they will use it well. Fewer than 10% of gen AI use cases make it past the pilot stage. It’s a matter of turning experimentation into value-creating acceleration. CEOs might accept that principle, but in practice they are nervous about getting it wrong. 

A critical way to get beyond the nerves is to build trust. This may sound fuzzy; it isn’t. Companies where gen AI usage accounts for at least 10% of earnings are committed trust-builders. They are also twice as likely to see high revenue growth (10% and up). In a survey of almost 1,400 companies, 43% of the gen AI high performers ensured that new models allowed audits, risk and bias assessment, compared to 18% for the rest. 

Four kinds of trust are essential. 

Trust in the data. If people don’t have confidence in gen AI’s output, they won’t use it, and there goes any chance of organizational transformation. Building this confidence needs to be done intentionally, thoughtfully, and thoroughly. That only rarely describes reality. No question: this isn’t easy, because there are two conflicting priorities. To create a trusted stock of knowledge that can be put to good gen AI use, the data needs to be transparent and accessible. At the same time, there needs to be effective management of data governance, security, and usage. Finding the right balance is a job for the C-suite. It is up to top leaders, with the support of an AI oversight committee and legal and regulatory experts, to define policies, establish compliance procedures, set up guardrails, and ensure that human judgment has the last word. 

Trust in gen AI’s decision-making capabilities. To deploy gen AI usefully, leaders have to answer a fundamental question: Where can we let it do most of the decision-making? The most promising applications of gen AI share common traits, including the reliance on digital information, the potential for scalability, and a foundation in standardized processes. 

Right now, these capabilities are at work automating routine, data-driven tasks, whether in back-office operations, document processing, or customer support. They enable personalization down to the individual level—adapting messages, services, or products in real time. They foster new forms of content creation and creative problem solving, such as generating code and designing new molecules. They accelerate decision-making, delivering insights from massive datasets faster than any team of analysts could, for example pricing recommendations in retail, risk assessments in finance, demand forecasts in supply chains, and diagnostic suggestions in healthcare. 

Such decisions play to gen AI’s strengths because they are rules-based, data-intensive, and repeatable. Most importantly, these examples demonstrate how gen AI creates value by amplifying the cognitive, creative, and decision-support capabilities that are the domain of people. 

Trust in its potential. This potential will not be the same everywhere; the rate of change will be uneven. The point is to pick the right spots. Parts of the organization could become minimum viable organizations (MVOs) where swarms of AI agents oversee most work, while people check their outputs. Other parts would rely more on people, working in tandem with gen AI and agentic AI. For example, in customer service, AI agents could resolve routine inquiries while people deal with the exceptions. In human resources, people would define roles and decide whom to hire but gen AI takes care of onboarding and payroll. This dual model highlights how organizations can flexibly combine AI-first units with human-led units, depending on the nature of the work, the need for trust, and the tolerance for risk. Leaders who embrace this vision will not only capture productivity gains but also begin to reinvent how their companies operate.

Advertisement

Trust in employees. Providing training reduces anxiety and builds confidence: the more skilled people are in gen AI, the more they use it. That starts with selecting the right work processes to automate first; this can increase buy-in because these improvements will make employees’  jobs easier. Employees should be invited to create their own agents and to suggest how to integrate gen AI into their workflows, making use of gen AI a habit, not a one-off. When McKinsey launched its internal gen AI platform, Lilli, for example, we made a point of asking in meetings whether Lilli has been involved; new employees are introduced to Lilli right away. The message is clear: Lilli is an essential, if invisible, team member.  

Ultimately, the opportunity is not only for people to do their job better but to become co-creators and active participants in how gen AI is used. To make this happen, companies can benefit from finding and trusting gen AI enthusiasts to become champions of change; these super-users can mentor their colleagues and demonstrate gen AI’s potential. 

Gen AI has the potential to completely change how companies operate and how people work—to everyone’s benefit. To reach that destination, trust is the North Star—one that illuminates how gen AI will make organizations, and their people, more successful. 

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.