Breaking Down the Barriers of Practical AI Adoption

It’s hard to find an agency today that isn’t aware of the benefits of artificial intelligence (AI). Many organizations are taking advantage of AI and machine learning (ML) to uncover deeper insights from data and act on them quicker than before.

With AI and ML, agencies can drive real, meaningful action with less resource-intensive involvement. That can result in anything from instituting better vehicle traffic patterns to making breakthroughs in medicine.

However, there are many barriers to entry that state and local governments encounter, such as data integration challenges and IT complexities. Even agencies that adopt AI often have trouble repeating their successes throughout the organization. How can agencies overcome these barriers and practically adopt AI for repeated success?

To learn more, GovLoop spoke to Daniel Domkowski, AI/ML Delivery Specialist at Red Hat.

Make Decisions Where the Data Sits

Imagine a clinician analyzing a patient’s MRI scans. It is far more resource-intensive to ship the MRI data to the cloud or data center than to conduct the analysis and get results right near the hospital room, or the edge location.

“Shipping all that data to a cloud environment, or to a centralized decision management location, is expensive. You lose time, resources and money that way, whereas having a consistent delivery pipeline across the hybrid cloud is far more cost-effective,” Domkowski said.

A consistent delivery pipeline and application platform can deliver results in any environment, whether the data is on premises, in the cloud, at an edge location, or at all three. The flexibility of this pipeline enables the system to execute tasks faster, analyze data and automate decisions in a setting like a hospital room rather than an off-site location.

Repeat Success With a Continuous Pipeline

Even when agencies can adopt an AI use case, how can another department share the success?

“There are ways in which you can scale success by following a consistent flow of work that is suitable for your entire organization,” Domkowski said.

In IT systems, agencies can identify and build a consistent flow of work through DevOps and Agile principles, which stimulate an iterative, incremental and highly interactive way of working. The principles also encourage capabilities to improve the application development process at all phases through monitoring and automation.

“If all your teams are able to operate on a consistent pattern in any environment, utilizing similar concepts and tools, with the flexibility to choose which tools to use, that can lead to repeated successes,” Domkowski said.

Treat AI Responsibly With Unbiased Data

Just because you’re able to deploy machine learning models doesn’t immediately mean you’re successful. The insights and modeling must be accurate, which means the data has to be clean and unbiased.

Bad data can lead to dire consequences, such as disrupted traffic patterns or unbalanced social services. Without quality data, AI-based decisions can lead to individuals not receiving the services they need or deserve, for instance.

“Treating machine learning and artificial intelligence with responsibility is critically important,” Domkowski said.

This article is an excerpt from GovLoop’s recent guide, “Resilience Lessons From State & Local Government.” Download the full guide here.

Leave a Comment

Leave a comment

Leave a Reply