Despite rapid advancement in the AI tech space, organizations are clearly struggling to turn implementation into scalable transformation. New research from the Deloitte AI Institute digs into the actions that lead to successful outcomes—providing leaders with a guide to overcome roadblocks and drive business results with AI.
The fifth edition of the firm’s State of AI in the Enterprise survey, conducted between April and May 2022, provides organizations with a roadmap to navigate lagging AI outcomes. Even though 79 percent of respondents say they’ve fully deployed three or more types of AI, because of advances in AI tech since last year’s report, 29 percent more respondents surveyed classify as underachievers this year.
“Amid unprecedented disruption in the global economy and society at large, it is clear today’s AI race is no longer about just adopting AI—but instead driving outcomes and unleashing the power of AI to transform business from the inside out,” said Costi Perricos, Deloitte Global AI and data leader, in a news release. “This year’s report provides a clear roadmap for business leaders looking to apply next-level human cognition and drive value at scale across their enterprise.”
The research outlines detailed recommendations for leaders to cultivate an AI-ready enterprise and improve outcomes for their AI efforts. Similar to last year, the firm grouped responding organizations into four profiles—Transformers, Pathseekers, Starters and Underachievers—based on how many types of AI applications they have deployed full-scale and the number of outcomes achieved to a high degree. The findings aim to help companies overcome deployment and adoption challenges to become AI-fueled organizations that realize value and drive transformational outcomes from AI.
“Since 2017, we have been tracking the advancement of AI as industries navigate the ‘Age of With,’” said Beena Ammanath, executive director of the Deloitte AI Institute, in the release. “The fifth edition of our annual report outlines how AI can propel businesses beyond automating processes for efficiency to redesigning work itself. While organizations face the challenge of middling results, it is clear successful AI transformation requires strong leadership and focused investment, a through-line consistently evident in our annual research.”
Four key actions powering widespread value from AI
Based on analysis of the behaviors and responses of high- and low-outcome organizations, the report identifies four key actions leaders can take now to improve outcomes for their AI efforts:
Action 1: Invest in leadership and culture
When it comes to successful AI deployment and adoption, leadership and culture matter. The workforce is increasingly optimistic, and leaders should do more to harness that optimism for culture change, establishing new ways of working to drive greater business results with AI.
- Eighty-two percent of respondents indicate employees believe that working with AI technologies will enhance their performance and job satisfaction.
- The highest performing respondents (“Transformers”) were the most likely to report AI-ready cultural characteristics, such as: high cross-organizational collaboration; workforce optimism for the possibilities of AI; and actively nurturing and retaining AI professionals.
- The survey found that agility and willingness to change, combined with executive leadership around a vision for how AI will be used, are the most important factors in the development of an AI-ready culture. Change management is critical to successful AI transformation, and high-outcome organizations were more than 55 percent more likely to invest in change management compared to low-outcome organizations.
- Organizations are taking action to support human-machine collaboration with 43 percent of respondents noting their organization has appointed a leader responsible for helping workers collaborate better with intelligent machines, and 44 percent say they are using AI to assist in decision-making at senior-most levels.
Action 2: Transform operations
An organization’s ability to build and deploy AI ethically and at scale depends on how well they have redesigned their operations to accommodate the unique demands of new technologies.
- In both the fourth and fifth editions of this survey, operational best practices were associated with high outcomes, but most organizations have yet to make significant improvement in this area. In both the fourth and fifth editions, just one-third of respondents say that their companies are always following best practices such as MLOps, redesigning workflows, and documenting AI model life cycles.
- Managing AI risk can have a major impact on an organization’s AI efforts, with 50 percent of respondents citing management of AI-related risks as one of the top inhibitors to starting and scaling AI projects.
- By and large, surveyed organizations rely heavily on training as a key to mitigating AI risk. Respondents’ top two risk mitigation strategies are training developers on AI ethics (35 percent) and training/supporting employees who work with AI (34 percent).
Action 3: Orchestrate tech and talent
Technology and talent acquisition are no longer separate. Organizations need to strategize their approach to AI based on the skillsets they have available, whether they derive from humans or pre-packaged solutions.
- Given that even the most advanced organizations are still early in their transformations, a majority of organizations still prioritize bringing new AI talent into the business from outside, rather than retraining existing workers (53 percent vs. 34 percent).
- A significant majority of the survey respondents acquire AI as a product or service (65 percent) rather than attempting to build their own AI solutions in-house (35 percent), leaning particularly on off-the-shelf solutions at the beginning of their journeys.
Action 4: Select use cases that accelerate outcomes
The report found that selecting the right use cases to fuel an organization’s AI journey depends largely on the value-drivers for the business based on sector and industry. Starting with use cases that are easier to achieve or have a faster or higher return on investment can create momentum for further investment and make it easier to drive internal cultural and organizational changes that accelerate the benefits of AI.
- The survey found the top use cases of AI across industries include cloud pricing optimization (44 percent); voice assistants, chatbots and conversational AI (41 percent); predictive maintenance (41 percent); and uptime/reliability optimization (41 percent).
- However, use cases vary by industry, for example:
- Life sciences and health care companies are the most likely to delegate ownership over AI models to individual lines of business (51 percent), while technology, media and telecom companies are most likely to centralize this ownership (42 percent).
- Energy, resources and industrials companies are most likely to use AI to assist in decision-making at the highest levels of the company (50 percent), while government is least likely to do so (39 percent).
The firm surveyed 2,620 executives from 13 countries across the globe.