Enthusiasm for artificial intelligence is running high, with early adopters ramping up their investments, reporting substantial returns, and planning to invest in more projects, according to Deloitte's second edition of the "State of AI in the Enterprise" report, released today. Nonetheless, organizations may want to tap the brakes – if only to more skillfully navigate the inevitable twists and turns that lie ahead.

In the survey of 1,100 U.S. executives from companies considered to be early AI adopters, a full 82 percent report a positive return on their investment, and the median return is 17 percent. The study reveals that some industries are more adept than others at turning investment into financial benefits. For example, the technology, media and entertainment, and telecommunications (TMT) companies are estimated to have a median ROI of 20 percent from their AI initiatives.

AI penetration is pervasive and growing amongst early adopters, with the most complex technologies having the highest adoption levels – machine learning (63 percent), natural language processing (62 percent), computer vision (57 percent), and deep learning neural networks (50 percent). The survey notes that advanced enterprise software is making it easier for companies to get started and benefit from AI, with 59 percent using enterprise software with AI baked in.

"Many of the complex challenges businesses need to solve today require humans working with machines to gain advantage," said David Rudini, principal and chief analytics officer, Deloitte Consulting LLP. "In order to achieve true ROI from your AI investments, it requires defining specific business outcomes, and understanding the costs, cascading impacts, and talent implications at the onset."

Cybersecurity tops AI concerns   
While survey respondents have several concerns with AI technologies, cybersecurity vulnerability is the biggest, outranking the fear of strategic business errors related to AI recommendations:

• Disturbingly, 32 percent of respondents have experienced an AI-related breach within the last two years.

• Thirty percent of the respondents have slowed an AI initiative in order to address cybersecurity concerns, and 1 in 5 have decided not to launch AI initiatives due to cybersecurity worries.

• Nearly 4 in 10 respondents indicate a high concern about the legal and regulatory risks associated with AI systems.

• Ethical risks are another worry, with 32 percent of respondents rating it a top-three concern. Ethical risks include the power of AI to spread false information, and chance of bias in AI algorithms that can skew recommendations in areas like lending or career recruitment.

Despite these issues, the survey notes that organizations can't afford to sit back and watch competitors extend their advantage. Sixty-three percent of respondents are using AI to catch up, keep up or "edge slightly ahead" of competitors, while 37 percent are using AI to notably widen their lead.

Getting AI right takes time
Nonetheless, companies realize that AI use is still maturing and that the window of opportunity remains open.

• Although 42 percent of respondents believe adopting AI will be of "critical strategic importance" two years from now, only 11 percent believe that's the case today.

• The share of respondents (56 percent) who believe AI will transform their company within three years, and the share (37 percent) who believe AI will transform their industry within three years, are both down 20 percentage points compared to last year's survey.

• Today's three most common AI uses are still largely IT-focused. However, the transformative potential of AI can only be reached when it permeates an entire organization.

"Companies are excited about the potential of AI to improve performance and competitiveness – and for good reason," said Dr. Jeff Loucks, executive director, Deloitte Center for Technology, Media and Telecommunications, Deloitte LLP. "But to reach this potential, companies must engage risk, address talent shortfalls and execute well. While AI's upside is significant, haste can leave companies with bridges to nowhere – pilots that don't scale or projects with no business benefit. The good news is Deloitte helps clients understand the pitfalls and how to avoid them."

AI's talent gap and job impact
Getting AI right also requires having the right talent. Many companies are feeling the squeeze in this area, with 69 percent of respondents facing a "moderate, major or extreme" skills gap. In addition to requiring talent, AI affects talent in the wider workforce. Early adopters are clearly searching for the right balance between using AI to reduce headcount and to augment the capabilities of their employees. While respondents don't view reducing headcount as a top benefit of AI, 63 percent have the perception that their company "wants to automate as many jobs as possible with AI/cognitive." This could be the recognition that automating menial tasks that are manual and repetitive in nature provides the ability to shift focus to higher value work.  

• As such, it follows that AI will affect job descriptions. A strong majority of respondents agree that AI leads to either moderate or substantial changes in job roles and skills both already (72 percent), and in three years (82 percent).

• For companies that have reported on AI skills gap, the most-needed talent is AI researchers to invent new kinds of algorithms and systems (30 percent ranked this a top-two need), followed by software developers to build AI systems (28 percent), and data scientists to manage and analyze the information that fuels AI (24 percent).

• Seventy-eight (78) percent of respondents agree that AI empowers people to make better decisions, and 72 percent believe that AI will increase job satisfaction.

• Executives are open to replacing affected workers. Only 10 percent of respondents cite a clear preference for retraining and keeping current employees. Seventy-eight (78) percent lean toward either "keeping or replacing employees in equal measure" or ''primarily replacing current employees with new talent."

Despite the potential job threats, the survey found that AI could blend the best of what machines do with human experience, judgment and empathy. Seventy-eight (78) percent of executives agreed that AI-based augmentation will fuel new ways of working. 

Toward execution excellence
Finally, Deloitte notes that getting AI right requires operational discipline and the pursuit of "execution excellence" in a variety of areas. Among early adopters:

• Fifty-four percent have a process for moving prototypes into production.

• Fifty-two percent have an implementation roadmap.

• Forty-five percent have appointed senior executives as AI champions.

• Thirty-seven percent have created a companywide center of excellence.

• Thirty-seven percent have a comprehensive AI strategy in place.