Making Shallow Adoption Deeper Through Trust
While AI adoption is rather broad across most civilian and defense agencies, current uptake is shallow. An overwhelming 85% of respondents indicate that potential security vulnerabilities are a concern for AI expansion, while others worry about potential bias or discrimination in AI outputs. Clearly, agency employees need greater assurances about the provenance of data, data governance, and how agents are using the appropriate context as part of their workflows.
“AI adoption only accelerates through trust,” added Purkey. “That means leaders and employees must share a common understanding of why AI is being inserted into their agency’s business processes, the accuracy and authenticity of the data agents will use to make decisions, and the right policies to ensure that human oversight is embedded in all critical workflows.”
Agencies continue to experiment with measures to increase trust and transparency in their AI adoption. Overall, 3 out of 5 respondents report that humans review some or most of their agentic workflows and outputs in their organizations.
Interestingly, nearly two thirds of respondents, representing large majorities in both civilian and defense agencies, indicate that executive leadership has the greatest influence over how agentic AI is deployed. With a foundation of trust, leaders are well positioned to make a significant leap in how they embed and embrace AI moving forward.