Alan Shark is an Associate Professor at George Mason University’s Schar School of Policy and Government, an affiliate faculty member at GMU’s Center for the Advancement of Human-Machine Partnership, and Executive Director of the Public Technology Institute.
He’s also the author of “Artificial Intelligence: A Primer for State & Local Government.” The cover of the book gives away his message: AI gets more capable every day, and governments need to learn about and experiment with it, starting now. The book was drafted, edited and illustrated with the aid of AI.
“I wanted to show proof of concept for what can be done,” Shark said.
Experiment Safely
The biggest AI mistake that state and local governments make is focusing solely on its hazards, he explained. “I think everybody is so afraid of what this could do. They’ve had these commissions and study groups and whatnot, and they’ve instilled a sense of fear,” he said. “What worries me is that everyone’s talking about guardrails and restricting use, coming up with things ‘thou shalt not do’ as opposed to what we should be doing.”
He recommends an atmosphere of curiosity. “Let’s experiment. Let’s see what we can do,” Shark said. One way to get users familiar with the possibilities is to create a sandbox where they can test implementations without releasing them to the wider system or to the public, he advised.
Shark also recommends overcoming fears with information. He’s laid out a sample curriculum for basic in-house AI training but says most agencies feel the need to bring in expert trainers. And because new AI uses emerge every day, he suggests having regular internal AI meetings for people to exchange their experiences.
A Valuable Tool
AI is already in use more than people realize, Shark said, especially in applications that improve individual productivity. Government enterprises have been slower than the private sector to adopt the technology. Where it is being used — for such as in digital assistants and chatbots — the quality of the responses has become more realistic. “The quality of voice responses has really improved, almost to the point where it’s hard to decipher whether this person is real or virtual,” he said. Data analysis offers additional opportunities.
“This is where AI works its best, where it can really do things that humans can’t possibly do within a given time frame,” said Shark. “An AI system can look at things and see anomalies, see patterns that we may very well miss, and can do it in microseconds.” That helps agencies make more informed decisions and predictions, and better understand health, public safety, transportation and other issues.
What About the Cost?
Although familiar AI offerings are free to the public, that obscures hefty costs, including energy and data center requirements. “We just type in this little dialogue box, or we download a document or ask it a question. We don’t see what goes on behind it,” Shark said.
Adapting the technology to agencies’ specific security, privacy and data access needs will be expensive, he explained, and agencies must make tough choices. Someone will need to decide which employees have access to AI applications, he said.
“Somebody has to say, ‘How does this meet our mission? How much better are the outcomes if we use this?’” Shark said. “You’re going to have to justify [it] in terms of costs [and] benefits. If the answer is, ‘It makes us more accurate, it’s more responsive to public need,’ then it becomes [easier] to justify.”
For most state and local agencies, AI will be available as a cloud-based subscription service. “You’re either paying for it on a monthly basis or on a usage basis, and you’ll have to understand up front what those costs will be.” As with other cloud services, it initially might be difficult to estimate usage. “It’s going to be a learning curve on two parts: recognizing [AI’s] cost and recognizing all it can do,” Shark said. “Not only are we going to use it and pay for it, we’re going to use it more. ” Enhancing, Not Replacing Shark acknowledged the concern that AI will replace jobs, but he said people and AI can work together to produce better results than either can alone.
“When [you] think of AI, most people think of it as ‘artificial intelligence,’” he said. “The AI that I think of today is ‘augmented intelligence.’ It’s augmenting humans in terms of making what we do better, more productive, and hopefully more accurate.”
This article appeared in our guide, “Getting Practical With AI.” To learn more about how agencies are using this important technology today, download it here:
Leave a Reply
You must be logged in to post a comment.