AI has revolutionized the business world, but it’s important to keep in mind that as robust as AI capabilities have become, the technology is still very much in its infancy. And at about the 18-month mark since generative AI began to be widely used, we’re at the stage where big changes, both infrastructural and regulatory, will more firmly shape the way the technology will be used in the future. New academic research from the University of Maryland’s Robert H. Smith School of Business takes the pulse of 885 US-based business executives and middle managers from for-profit companies, revealing both the concerns and support surrounding AI adoption and governance.
One key finding of the new report, AI Use and Regulation: A Survey of U.S. Business Executives, is that with governments now formulating strategies for AI regulations, it will be critical for businesses to successfully adapt to AI policies as they emerge, said Kislaya Prasad, research professor at the Smith School who conducted the research, and academic director of the school’s Center for Global Business, in a news release.
The report begins with five key takeaways:
- Considerable concern exists about job displacement and is foremost in financial services and insurance and telecommunications.
- Strong support is evident for AI regulation, including mandates for transparency about AI use, explaining autonomous decisions and undergoing third-party auditing for bias in algorithms.
- Strong support exists for restrictions on export of key AI technologies.
- “Powering chatbots” and “coding” are identified as the most important uses for Generative AI, which was already widely used across sectors by November 2023.
- While “improving customer experience” and “improving operations” are key drivers of AI adoption, major reasons for non-adoption are an “absence of a clear use case or perceived need” and “limited technical expertise of resources.”
Collectively, almost 58 percent of respondents reported that their firms had incorporated AI into their business practices in some capacity, 35 percent said they had not, while the remaining 7 percent stated they were unsure about the level of AI integration at their company.
The report further addresses job displacement, support level for AI regulation and export restrictions, sentiments on the patentability of AI-assisted creations and intellectual property infringement, AI use by sector, and the drivers and hurdles associated with AI adoption.
A strong backing for AI regulations exists
The Biden Administration’s 2023 Executive Order on AI aimed to establish new standards for AI safety and security, create privacy safeguards and promote innovation and competition in business. Over the past five years, 17 states have enacted 29 bills on AI regulation promoting similar principles. As for the extent of support among executives for regulation of AI-based systems, respondents were asked about three types of mandates—transparency about AI use and data collection, explainability of autonomous decisions by AI algorithms and third-party auditing for the presence of algorithmic bias in AI algorithms. Approximately 75 percent of responses declare to strongly or somewhat support regulation mandating transparency, with algorithmic bias regulation held in a similar regard. About 72 percent of respondents strongly or somewhat support explainability regulations.
Job displacement concerns weigh heavily on executives
Regarding the potential adverse impact of AI on career prospects over the next five years, roughly 20 percent of respondents expressed that they were either very or extremely concerned. These worries resonated with 47 percent of participants from the financial services and insurance sector, and with 32 percent in telecommunications. Additionally, 27.5 percent of respondents with less than 15 years of work experience and 26 percent of respondents who identified themselves as AI decision-makers at their respective companies share this concern. Although there is discernible concern among people directly involved with AI in their work, “it’s not clear if this stems from more intimate knowledge of AI’s possibilities or from being in more vulnerable roles,” writes Prasad.
There is resounding support for restrictions on exporting key AI technologies
In addition to the 2023 Executive Order on AI, the U.S. Department of Commerce strengthened export controls on AI technology, targeting the sales of advanced chips and chip making equipment to China. According to Secretary Gina Raimondo, the goal was to limit China’s “access to advanced semiconductors that could fuel breakthroughs in artificial intelligence.”
Support for those policies was apparent among survey respondents, with almost 60 percent strongly or somewhat supporting restrictions. Firms with 10 percent or greater international sales supported AI technology export restrictions more significantly. Manufacturing led all sectors by a sizable margin, with 70 percent of its respondents strongly or somewhat supporting restrictions on exports of cutting-edge AI technology. Older respondents, people concerned about AI-related job displacement and those with high trust in the government are more likely to support export restrictions, too.
Generative AI has the early lead in AI adoption within business
When asked about the AI technologies implemented by their companies, 39 percent shared that generative AI, followed by computer vision (30 percent) and machine learning (27 percent), were in use. Firms with a significant global presence proved to be the most intensive users of AI for generative tasks. Among respondents from these firms, 33 percent said they used generative AI for chatbots, while 32 percent used it for marketing purposes and 30 percent for text generation. Regarding the decision-making tasks that currently use an autonomous decision system, respondents regularly cited inventory management, logistics, personalization and recruiting.
Customer experience and operations efficiency improvements are at the core of AI adoption
Drivers and hurdles, overall, were similar across sectors. However, at companies where AI is in use, those two drivers appeared among 66 percent and 72 percent of responses, respectively. Hurdles selected by more than 35 percent of firms with adopted AI technologies included high initial costs, difficulty recruiting skilled professionals and the challenge of integrating AI with existing IT infrastructure. As for companies where AI technology was not adopted, the two most cited reasons were the absence of a clear use case or perceived need for the technology and limited technical expertise or resources to implement and manage the technology.
“There is great similarity in patterns of use of AI across sectors, although levels vary widely. Information technology, telecommunications, financial services and insurance, and manufacturing have much higher levels of AI use than, say, retail and e-commerce,” said Prasad.
However, AI is being used in similar ways everywhere, he adds. “Moreover, sentiments towards AI and its regulation are similar across sectors.”
Funding from the U.S. Department of Education through a Title VI grant under the CIBE program contributed to this research. Survey respondents were chosen primarily based on the ability to provide diverse responses and viewpoints on AI implementation across industries. Respondents spanned eight sectors comprising roughly half of the U.S. private sector workforce: financial services and insurance, healthcare and biotechnology, hospitality and leisure, information technology, manufacturing, retail trade, telecommunications and transportation. A ninth category, “Other,” was included to represent individuals outside the eight main sectors.
