US Environmental Protection Agency CIO Carter Farmer is blunt in his message to AI hype-chasers. Shiny object syndrome leads teams to jump into AI without defining clear use cases or vetting data, and then wonder why it does not work.
Farmer, speaking during a webinar organized by FedInsider said that AI wasn’t the magic cure-all for business operations as many people believe. The EPA CIO stated that AI must be introduced into a company with a specific application in mind.
“People hear AI and they think it can solve any problem,” Farmer argued. “They see what it’s done at other organizations and jump in without asking the right questions.”
The problem you are trying to solve may not require AI
he explained “Many times the problem you’re trying to solve doesn’t need AI,” he added that jumping on the latest buzzwords could slow down growth rather than accelerate it. Farmer said.
This is a reasonable view, considering that investment returns in AI have been pretty poor recently. According to a survey of 2,000 CEOs, only one out of four AI bets pays off . Farmer’s comment was in response to Ed Bodensiek who is the customer experience market leader for government tech services firm Maximus.
Bodensiek stated that he has heard other government technology leaders say there are too many robots in government agencies. His company has developed a “mission readiness assessment” method to determine what the best approach is to solve a problem, which may not involve AI at all.
What’s wrong with so many AI projects? Farmer says it’s two things: Data and processes. Farmer explained that applying AI to business processes does not mean mapping what you do now onto an AI.
- AI investment may be an $8T gamble
- AI agents perform office tasks incorrectly around 70% of the times, and many of them aren’t AI at all.
- Wearers of Tinfoil Hats can thank AI for the declassification JFK documents
- IT decision-makers are not convinced when it comes to AI ROI
“What people should be doing is redefining and reimagining that process as it would apply to automation,” said the CIO.”[The] lift and shift mentality has to be stopped in its tracks.”
Farmer gave an example of a hypothetical situation in which a form, which had been part a business process for many years, was still being filled out even though it was now performed automatically and digitally. It’s possible that a form could be included in an AI workflow without the business process being examined.
A review of the process can force teams to ask critical questions, such as whether AI-ifying something is worth the effort. “Value add might come at 2x, 3x or 4x the cost of doing it manually,” Farmer explained.
To summarize, if AI can only be as good or as bad as the data it is based on, then an AI project can only be as good or as bad as the data it was based on.
Farmer’s work has been shown to that end. The EPA maintains an inventory of AI applications at the agency as part of his philosophy “finding the actual best use cases where AI can be best applied,” which he explained in the webinar. (r)
