In 2024, organizations gained experience with Generative AI (GenAI) and better understood the rewards and challenges of deploying it at scale. As C-suites and boards made efforts to adopt AI, the need for discipline is essential. While many questions about GenAI have been answered, many remain, with others just beginning to be proposed.

For the past year, Deloitte conducted a quarterly global survey of executives that focused on GenAI in their enterprise. It’s given us a closer look into how companies are keeping pace with this fast-paced technological environment, aiming to stay true to their brand and continue to fulfill customer needs.

For our fourth wave report, we sought to answer several questions around value realization, investment and scaling to help build a roadmap for the future.

Where does workforce adoption stand? Our latest survey results show that access to GenAI is still largely limited to less than 40% of the workforce. And for most organizations, fewer than 60% of workers who have access to GenAI use it daily, which suggests many companies have yet to integrate it into their workflows. For GenAI to become truly transformational, it will require far greater experimentation and identifying new, high impact use cases.

But access alone does not equate to success. Providing access to GenAI does not mean workers will use it. Conversely, workers with a desire to use GenAI will find a way to do so, with or without approval. To foster transformation and maintain a level of control over how GenAI is used within the enterprise, it generally makes sense to offer broad workforce access to sanctioned GenAI tools, with clear guidelines for proper use. We also see this as a pressure point in most organizations as the desire to release broad-based GenAI tools comes with a cost and the efficiency benefits are hard to measure and/or small portions of an individual's time of their daily activities.

What are the success rates of experiments? We found organizations are still heavily experimenting with GenAI, and scaling tends to be a longer-term goal. Over two-thirds of respondents said that 30% or fewer of their current experiments will be fully scaled within the next three to six months. This suggests companies are taking time to test GenAI capabilities and determine where it can help the most.

As expected, actions across the market vary by organization, with larger companies experimenting more, and by industry/sector, depending on access to data and regulation – i.e. we’re seeing financial service organizations applying AI quicker than, say, government services.

What use cases are showing promise? To understand where GenAI is having the deepest impact on organizations, we asked respondents to consider one of their most advanced GenAI initiatives. Since GenAI is a highly advanced technology—and one of its best capabilities is generating computer code—it’s no surprise that the IT function came out on top (28%). However, the survey data also shows GenAI being deployed deeply in many other parts of the business as well, including operations (11%), marketing (10%) and customer service (8%).

Even more revealing, we found that the most advanced GenAI applications outside of IT target critical business areas fundamental to success in a company’s industry (e.g., marketing in the consumer industry; operations in energy, resources and industrial; cybersecurity in financial services). This is a crucial insight since many business leaders still associate GenAI with personal productivity and other relatively mundane tasks secondary to the core business.

Is GenAI meeting ROI expectations? Return on investment (ROI) for organizations’ most advanced GenAI initiatives has been generally positive, with almost all organizations reporting measurable ROI. Despite longer than expected time to value, nearly three-quarters of respondents reported that their most advanced GenAI initiative is meeting or exceeding ROI expectations.

Cybersecurity implementations have emerged as a standout, with 44% saying ROI surpassed their expectations, more than any other function. These results are somewhat skewed by advanced GenAI deployments in the financial services and technology industries, where cybersecurity is especially critical. However, the relatively strong performance of cyber-related GenAI initiatives makes sense for other reasons as well, as many organizations are already experienced in using AI to manage cyberthreats and have related infrastructure in place to scale cyber capabilities.

What does the future look like? Among all the emerging GenAI-related technological innovations, agentic AI appears to be capturing the most interest and attention. The vision for agentic AI is that autonomous AI agents can execute assigned tasks consistently and reliably by acquiring and processing multimodal data, using various tools to complete tasks and coordinate with other AI agents—all while remembering what they’ve done in the past and learning from their experience. Agentic AI is the next logical step for GenAI, giving GenAI-based systems access to more types of information and increasing AI’s level of responsibility and autonomy.

As with previous transformational technologies, the initial excitement and hype about GenAI have gradually evolved into a mindset of positive pragmatism. Many companies are already witnessing encouraging returns on their early GenAI investments. However, these companies and others have learned that creating value with GenAI—and deploying it at scale—is hard work. Although the technology can seem magical, there is no magic wand for GenAI adoption, deployment, integration, and value creation -- it requires strategic planning, dedicated effort, and continuous innovation.