Artificial intelligence might be a disruptive technology, but it’s not magic.

To deploy AI effectively requires knowledge and training, together with a strong sense of how to use it to enhance missions, advised Drew Firment, chief cloud strategist at Pluralsight. In that sense, it resembles cloud computing, which agencies have learned to normalize over the last decade.

“There are a lot of lessons, I think, from the consumption and adoption of cloud computing that apply directly to AI for agencies,” Firment said during Federal News Network’s Industry Exchange Cloud 2024.

Chief among the similarities lies the need to create value with AI. “As folks move from tactical consumption of cloud services to really trying to create value with cloud services, same idea with AI,” he said. “It’s not just a matter of going out there and consuming AI as a science project, but rather how we create value.”

Value can take the form of improved services to constituents or outward-facing applications. It can also help improve internal operations, Firment said. The former use case is riskier because of concerns about privacy and bias that come with AI. So it might be wise to start with internal efficiency applications, he recommended.

In fact, AI can help with cloud operations themselves, Firment said. Optimizing costs, planning workloads and virtual machine capacity planning can all be revved up with AI, he said.

“Monitoring is really where we’re going to start seeing the use of AI — the aggregation and summary of all those logs and presenting that in an intelligent way,” he said. More than sorting out the real threats from the noise, an AI program can make recommendations on actions to take and even initiate remediation.

Firment said one way to keep cloud efficiency in check is to concentrate efforts on a single major provider.

“Pick a cloud and get good at it,” he said. “Don’t dilute your value proposition by trying to use all three [major vendors] and worrying about vendor lock-in.” Firment said the three majors are also rapidly developing their AI services.

“It’s just sort of pick one and get going, in my opinion,” he said. As example, he noted Amazon Web Services’ Bedrock offering. Like the AI offerings of other major cloud providers, it gives agencies ready access to large language model applications.

Using Bedrock or Google’s AI platform means “you’re not so hot and bothered to create that ecosystem” from scratch.

What cloud AI looks like

Set practical expectations for cloud AI use, Firment advised. For example, when trying to improve customer experience, “it could be as simple as chatbots and how to make them more intelligent.”

Or when working with large language models, run them on internal policies or procedures, he suggested. “When folks are interacting with you, they’re getting real answers, not just ‘listicles’ of your procedures.”

For internal applications, Firment reiterated the value in applying AI to cloud operations.

“There’s a big opportunity to help improve the efficiency of our consumption of cloud, which ultimately can reduce costs,” he said. Another opportunity would replace intranets with internally focused, intelligent bots to give employees access to documentation and policies.

“That might be a low-cost way to start paving the way and starting to learn some of the basic skills required,” Firment said.

However an agency approaches AI projects, he said to expect 20% of costs to come from the technology itself and 80% to come from organizational change and setting up use governance.

Don’t limit AI training to just the technologists, he added. “Education is going to start with leaders within the agencies — and even nontechnologists — to understand the language of AI and cloud. It’s fundamentally important to have that basic literacy.”

Discover more articles and videos now on Federal News Network’s Industry Exchange Cloud 2024 event page.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.