Federal IT leaders seeking to expand the government’s use of artificial intelligence (AI) are encountering significant workforce challenges, with the vast majority saying their technology teams lack critical AI skills, according to new MeriTalk research.
In a survey of 150 Federal technology decision-makers familiar with AI, MeriTalk found that 87 percent say their organization’s in-house workforce has only a fraction of the knowledge needed for AI innovation. Half of Federal organizations reported having an AI project fail due to a lack of expertise to support it, resulting in the vast majority depending on the private sector to augment AI expertise, the study showed.
But the research also illustrates AI’s potential as much as its challenges. Nearly all Federal officials surveyed feel that appropriate use of AI could supercharge the effectiveness of government, and 95 percent are currently investing in in-house AI skills development.
Bob Venero, CEO of Future Tech Enterprise, Inc., a global IT solutions firm, says the results demonstrate the need for public-private partnerships to ensure Federal AI success. “Successful AI requires a blend of attracting and retaining key talent, ongoing training, investing in the latest AI-powered hardware and software solutions, and making sure you have a business-focused AI strategy,” says Venero, whose firm focuses on serving Federal system integrators (FSIs).
Michael Shepherd, a senior distinguished engineer at Dell Technologies, says increased adoption of AI represents a “tremendous amount of opportunity” for Federal agencies, despite the workforce challenges.
Investing in AI “is going to make a difference,” Shepherd said in a recent interview with MeriTV. “I guarantee you, it’s happening in other countries, and we need to have that same level of investment here in the U.S. as well, especially within the armed forces and the Federal government.”
The MeriTalk research comes as interest is growing in the application of AI for Federal cyber use cases, and as AI experts and government officials focus on the importance of better preparing workers for the emerging technology.
The National Security Commission on Artificial Intelligence, for example, recommended in its final report last year that the government invest in upskilling the Federal AI workforce with educational courses and other training. The U.S. Senate in December passed legislation that would create a training program to improve the Federal workforce’s understanding of AI. A companion bill is pending in the House.
The MeriTalk survey, conducted in partnership with Future Tech, sought to better understand how Federal agencies are meeting AI staffing needs. The results identified critical resource gaps and the skills most in demand to fill them.
The biggest resource gaps include technical expertise, training on AI application and data management skills, and training on navigating trustworthy AI.
The AI experiences most in demand are, in order of importance: DevOps; data visualization; advanced statistics/modeling; responsible AI/algorithm bias; coding or programming; data literacy; and data mining or interpretation.
To fill the current skills gap for in-house AI expertise, 63 percent of respondents said at least half of their AI work is done by external vendors, FSIs, or contractors. Nearly three-quarters of organizations expect to increase their use of outside help as their AI strategies advance.
Among specific agencies, the Department of Defense (DoD) stood out. Compared to its civilian counterparts, DoD organizations were significantly more likely to say at least half of their AI work is done by contractors, by a margin of 73 percent to 53 percent. At the same time, DoD organizations are significantly less likely to have an AI project fail due to due to lack of expertise, with a 35 percent DoD failure rate in contrast to 63 percent for civilian agencies.
The other findings include:
- The seven most important factors for successful AI implementation, in order of importance, are: in-house expertise; technology; data preparedness; support from FSIs or other contractors; support of vendors; organizational culture; and luck or other factors
- Yet less than half of those surveyed feel fully prepared in any of those seven categories, with preparedness levels ranging from 28 percent for luck and other factors to 41 percent for in-house expertise
The report concludes with a number of recommendations for Federal contractors seeking to work with agencies on bringing the benefits of AI to the government. Among them: contractors should work with government IT leaders to bring senior-level decision makers to the table and double down on civilian partner engagements, while also developing their own workforces in the areas that agencies need most.