Carter Farmer, CIO at the US Environmental Protection Agency, has shared a critical perspective on the prevalent issue in AI project failures—rash decisions driven by ‘shiny-object syndrome.’ Farmer emphasizes the importance of having a defined purpose and data readiness before diving into AI implementations, which is often neglected in the race to adopt new technology.

During a FedInsider webinar, Farmer remarked that AI should not be seen as a magic fix for business operations. Instead, organizations need to initiate AI projects with specific intentions and explore the right questions early on. Missteps in the early stages can lead to significant setbacks.

Farmer further highlighted that many business problems might not require AI solutions. Frequently, the attempt to incorporate AI based on current trends without thoughtful deliberation can hinder progress. Reflecting on recent surveys, it’s evident that a substantial number of AI initiatives fail to deliver returns, as many organizations are hasty in their implementation without thorough consideration of data and processes involved.

In conversation with Ed Bodensiek of Maximus, Farmer discussed how government agencies mistakenly deploy bots, lacking a strategic approach. Bodensiek’s firm uses assessments to understand whether AI is the best fit for a particular challenge.

Farmer insists that strategic reviews can reveal whether the application of AI is truly beneficial. Often, organizations need to reimagine their processes rather than directly applying AI to existing structures. Moreover, the value derived from AI should outweigh its costs—sometimes manual methods are more effective. With AI only as effective as its input data, robust planning is crucial for any project.

The EPA maintains an inventory of AI implementations, illustrating Farmer’s advocacy for strategic application where AI provides genuine advantages.