There are lots of ways to think about generative artificial intelligence—how to use it, where to deploy it, and how to gain the most value from it. The answers to these questions are largely dependent on the industry or sector asking them. To better understand how approaches to gen AI vary by industry and how organizations can gain an industry advantage, we examined findings from Deloitte’s third quarter 2024 State of Generative AI in the Enterprise survey.
The data illuminated three key areas of opportunity for organizations to consider: business strategy adjustments, industry-specific use case alignment, and leveraging ecosystem capabilities.
Most organizations plan to increase investments in gen AI, according to the survey. At the same time, many companies are getting stuck in proofs of concept, with 70% of respondents saying that fewer than a third of their projects have moved to production, with energy, resources, and industrial respondents having scaled gen AI programs the least of the industries surveyed (figure 1).
One reason for this may be that companies haven’t connected their gen AI use cases closely enough to their overarching business strategy, focusing instead on adopting gen AI as quickly as possible rather than tying it to areas where they are trying to drive specific outcomes. This is not uncommon when a new technology comes on the scene; many will rush to invest in technology for technology’s sake, rather than considering where and how it fits into the broader business goals.
In doing so, they may be stuck in a proof-of-concept stage because they have not linked their pilots to clear, differentiated business outcomes. Indeed, most respondents note that they are not yet tracking return on investment, with life sciences and health care respondents ahead of the curve and energy, resources and industrials respondents lagging other industries in measurement strategies. Additionally, only 40% or fewer of all executives surveyed believe their organization has a high level of gen AI expertise, with technology, media, and telecommunications respondents reporting the highest expertise levels.
While we’ve explored the benefits of measuring the value of tech investments more broadly, it can be particularly valuable to do so for gen AI, given the technology’s nascence and the need to set benchmarks, understand progress achieved, and determine what proofs of concept, if taken to production and adopted, will return appropriate value to justify the effort involved.
This isn’t necessarily a problem. Rather, this could suggest that leaders still have significant opportunities to gain gen AI expertise, address risks, build data capabilities, and create ROI measurement programs that scale the impact of these initiatives and traditional AI initiatives overall. Leaders can benefit from integrating ways to track these key performance indicators into the solutions themselves, so it is easy to clearly and consistently articulate ROI by design rather than as an afterthought. Each of these areas—across people, data, risks, and outcomes—is important to a fully formed business strategy for effective gen AI implementation. Given the early stages of gen AI, organizations may want to take these steps now.
Even relatively more advanced industries—such as technology, media and telecommunications—still have headroom, as well as potential pitfalls to consider. Given above-average investment levels, they could potentially face even greater pressure to maintain their leadership position.
Our survey shows that the level of gen AI adoption across business functions appears to align with industry-specific value drivers. Consumer industry respondents are more likely to cite progress in marketing, sales, and customer service; financial services institutions are more likely to claim progress in finance and legal, risk, and compliance; life sciences and health care organizations are more likely to focus on product development and research and development. Similarly, tech and media technology respondents are more likely to be investing in strategy and operations and information technology and cybersecurity—a benefit not just for their organization but for the broader ecosystem as well (figure 2).
This suggests leaders are focusing gen AI attention and investment in functional areas of the business where they may already have deeper domain expertise and capabilities. That focal point is understandable, but it also suggests underutilization of gen AI capabilities in areas that may better align with the business strategy. This illuminates a significant opportunity for organizations across all industries to harness gen AI for greater value.
Here, a deeper look at financial services respondents’ prioritized benefits and impact can be instructive.
Seventy percent of financial services respondents report that they are increasing gen AI investment given the strong value they’ve achieved to date. The top benefits they expect from gen AI investments are improved efficiency and productivity, enhanced customer relationships, and innovation and growth (figure 3).
Moreover, our research shows that financial services industry leaders are deploying gen AI in areas such as fraud analysis detection; customer insights and experience; pricing; and information risk, compliance, and security. For example, major banks have made public investments in chatbots to aid financial advisers in making next-best-action customer recommendations, synthesizing investment research, and automating code for payment communications. In interviews conducted for this research, we heard the same: “Where you have more volume and data is a natural place to focus, and for us, that is retail and commercial banking,” says Dr. Yannick Lallement, chief artificial intelligence officer at Scotiabank. “We use gen AI to review client communications and create the appropriate use cases in our case management system, saving hours of total fulfillment time, which optimizes employee efficiency and results in a better customer experience.”1
According to our analysis, compared to other industries, fewer financial services respondents are struggling to define and measure gen AI’s impacts. This could be because they are above average in their efforts to directly tie gen AI investments to profit and loss and are more likely than other industries to have built a framework to evaluate gen AI investments. There’s evidence that financial services respondents’ approach to tracking ROI is connected to relatively stronger gen AI outcomes. For example, not only are they the only industry that says they have achieved enhanced relationships with customers as a top three outcome of their gen AI investments, but also they are more likely than the average to track nonfinancial benefits, such as relationships, as part of their approach to tracking value created.
As one leader from a global insurance organization that was interviewed for the State of Generative AI in the Enterprise study explains, “In terms of our objectives, everything we do in AI targets four outcomes: client experience, colleague experience (productivity), margin, modernization. We measure metrics like invoice error rate and effective tool use.”2 These findings can be instructive to organizations in other industries, many of whom have not yet begun evaluating or tracking investments to the same extent.
Financial services respondents report that embedding gen AI into functions and processes is the most valuable action for driving gen AI initiatives within their organizations, with 24% of respondents highlighting its importance. This priority significantly surpasses the average value of 11 percentage points, suggesting a solid consensus on its critical role and a belief that gen AI’s true potential is realized when it is seamlessly woven into the fabric of organizational operations (figure 4).
Gen AI model usage is currently all over the map as vendor ecosystems and capabilities rapidly mature. It can be difficult to wade through the range of options. The majority of industry leaders report investing in two primary ecosystem capabilities: productivity applications with integrated gen AI and publicly available large language models (LLMs). Other tools are leveraged to a much lesser extent—and some, such as customized LLMs or proprietary LLMs, hardly at all.
Interestingly, a relatively lower percentage of respondents are using enterprise platforms with gen AI already embedded—applications that could enable broader, enterprise wide adoption of gen AI capabilities in regular tasks. This suggests an opportunity for organizations to better utilize their ecosystem partners and explore broader use of gen AI-enabled tools.
At the same time, some industries are finding industry-specific applications and tools to provide an advantage: life sciences and health care, for example, places a much higher focus on industry-specific software applications, while tech and media technology executives are more likely to report using gen AI code generators. This industry-specific focus on tools flows in the other direction as well; likely owing to strict regulations around privacy, life sciences and health care executives are less likely to report using publicly available LLMs (figure 5).
While the gen AI technology ecosystem is still developing, and there are many unknowns, organizations can start with business outcomes in mind and work backward to wade through the options. Our analysis shows the top three outcomes prioritized by respondents from each industry (figure 6).
And there will be many decision points: model selection, application integration, human-in-the-loop coding approaches, open-source bets, and how the organization chooses to engage with providers and across the ecosystem can open up (or close off) the organization to future opportunities across its broader data, AI, and digital strategies. Knowing what an organization wants to achieve in the end can help narrow the choices to those that can produce the best results. Beyond the technological architecture, leaders can also tap into expertise and capabilities across their ecosystem to address talent gaps, enhance maturity, and achieve greater scale across their gen AI solutions.
At the same time, the data challenges and risk and governance concerns expressed earlier also should be addressed as part of that ecosystem strategy. Indeed, creating strong data foundations, better managing ecosystem privacy and security challenges, and reducing technical complexity across the digital ecosystem can create more value and drive industry advantage.
The transformative potential of gen AI is evident across all industries, offering unprecedented opportunities for growth, innovation, and competitive advantage (figure 7).
As organizations navigate the complexities of adoption, they can focus on ways in which gen AI can advance their tech strategies (rather than simply adopting it for adoption’s sake); look for opportunities to build gen AI into commercial, product, and operation functions; and leverage the power of the ecosystem. The journey toward gen AI maturity may be intricate, but the rewards are expected to redefine the future of AI, the organization, and the industry.