The journey ahead for AI adoption
The government stands on the brink of a new world of technology systems systems empowered by artificial intelligence and machine learning tools.
Such a shift could make government operations more efficient and effective. Our lives could be easier and richer.
That’s the hyped promise of AI and that vision might eventually come true. But while we may be on the brink, it will still take some effort for AI to cross the threshold into the mainstream.
A new study by the Professional Services Council Foundation (PSC’s research arm) lays out several of the challenges as well as the opportunities that AI adoption creates in the market.
PSC estimates that spending on AI will reach $1 billion, so no one doubts that “AI is here and it is here to stay” as PSC President David Berteau put it.
The association looked at AI from the government perspective and interviewed a wide range of government officials including chief technology officers, chief data officers, and several officials running AI projects at agencies.
PSC's report lays out seven areas where AI will have the most impact:
- Reduce administrative costs and reduce manual burdens, freeing employees for more valuable work.
- Optimize resource allocation
- Accelerate and improve decision making
- Reduce or eliminate backlogs and address increasing workloads
- Extract value from data
- Combat inefficiency and fraud
- Improve programmatic performance
The report also cites existing government use cases where AI is being applied. For example, the Health and Human Services Department is using AI to find ways to ways to consolidate contract vehicles.
“HHS is using AI to scrub purchase orders to find savings and eliminate duplication,” PSC Executive Vice President and General Counsel Alan Chvotkin said.
Other examples of current AI projects include the General Services Administration, which is using robotic process bots to track invoicing and payments. The Defense Department is using AI to plan deployments. And NASA is using bots to aid in finance and procurement process.
So AI is gaining some traction but there are challenges and some of those are more cultural than technical.
The government needs to write contracts that focus more on outcomes rather than merely buying inputs, Berteau said.
But to get there, he said the government needs a deeper understanding of AI and what it adds to a project and the technical and ethical implications.
Right now the government doesn’t broadly have the expertise to evaluate AI-rich proposals, Berteau said.
But industry has a role here in educating the government about AI and how it evaluate it and buy it, said Dominic Delmolio, managing director and CTO for Accenture Federal Services. Accenture was the underwriter of the PSC research report.
Delmolio said that industry also is increasing its use of AI as part of the solutions they submit to governments. But here again the challenge is whether the customer has the expertise evaluate the AI portion of the proposals.
The PSC report also lays out several policy initiatives that should help foster greater use of AI.
The Modernizing Government Technology Act requires that funds be repaid. So AI should find favor because of the return on investment AI empowers.
The March 2018 President’s Management Agenda outlines many cross-agency performance goals and detailed progress reports on the goals. The need for more data analysis should encourage more AI.
Bipartisan interest in AI could help as well, along with AI-enabled procurement to find costs savings and reduce duplication.
The increasing use of Other Transaction Authority contracts encourages the use of emerging technologies such as AI. A White House executive order on maintaining U.S. leadership in AI calls for sustained investment and development of AI.
The government is still in the “crawl and walk” stages of AI adoption. But it is moving toward the “run” phase. To get there, a wide range of government constituents -- senior leaders, managers and the general workforce -- need to become comfortable with the technology.
As one government official told PSC: “The biggest challenges have been the people challenges.”
Transparency and understanding how AI reaches its decisions is critical.
“How did you get there? Why did it arrive at that? It’s not just about delivering the end product – it's about having trust in that answer because you’re going to have the technical people challenging it,” another official told PSC.
Posted by Nick Wakeman on May 23, 2019 at 11:01 AM