The pandemic has put governments around the world to the test. To cope with increased demands, services have needed to be bigger, faster and stronger — bigger in availability, faster in handling requests, and stronger in the face of cyberattacks and network strain.
“We’ve seen a huge increase in artificial intelligence being put into production, including supporting mission-critical activities,” said Joe Pringle, Tech Business Development Manager of Artificial Intelligence (AI) and Machine Learning (ML) for Worldwide Public Sector at Amazon Web Services (AWS).
Cloud services have gotten the job done. They’ve allowed agencies to quickly and securely spin up AI projects across a spectrum of use cases, from education to health care. Those projects have real, tangible benefits for constituents, like more individualized curriculums and more accurate diagnoses.
GovLoop recently asked Pringle about the steps agencies should take to elevate AI using the cloud.
The benefits of the cloud begin when you launch AI projects. Delivered by the cloud provider, managed services for AI-enabled computer vision, speech, predictions and search have industry partners assume the onus of training, deploying and maintaining underlying models. Agencies’ biggest responsibility is developing use cases.
Moreover, once services are online, agencies don’t have to worry about overloads and crashes, because they can scale up or down to meet demand.
For agencies, cloud managed services have come in handy when they’ve needed to launch innovative programs in response to the pandemic. Those have included using computer vision to detect objects in videos, such as recognizing whether people wore masks in crowded settings.
Cloud-based AI can help agencies move faster, too. During the pandemic, it has. “We’re seeing a lot of governments using AI to automate highly repetitive and time-consuming tasks,” Pringle said.
One example is automating document workflows so that AI replaces manual data entry and extracts metadata to enhance search capabilities. As a result, AI speeds up timelines for constituents. Without having to wait on employees to manually enter data or respond to simple queries, citizens receive the front-facing information and services they need faster.
Agencies can build AI faster in the cloud, too. Developers access capabilities through simple application programming channels, so they don’t have to build or integrate models from scratch. Cloud services like Amazon SageMaker remove the busywork and infrastructure so that data science teams are more productive and efficient when rolling out ML.
In government, developing models quickly and broadly is useless if they fail the necessary security and compliance checks. AWS can step in at any point to shore up the ML pipeline so that the whole process — from data acquisition to service delivery — is secure.
“Security and compliance are critical at each and every stage of a machine learning pipeline,” Pringle said.
To maximize success, Pringle suggested three steps for launching a project. First, gain familiarity with how you can use AI. Then, build in-house capabilities and expertise, both business and technical, to vet vendors and assess offerings. Finally, start with a small proof of concept with clear value and minimal risk.
This article is an excerpt from GovLoop’s recent guide, “The State of AI in Government: Policies, Challenges & Practical Use Cases.” Download the full guide here.