Insight by Three Wire Systems

How agencies can embrace intelligent automation

With the exception of the military, where staying ahead of adversaries is an existential imperative, the federal government largely doesn’t have to worry abou...

This content is provided by Three Wire Systems.

With the exception of the military, where staying ahead of adversaries is an existential imperative, the federal government largely doesn’t have to worry about competition. The unfortunate result of this can be situations where the citizens are technologically far ahead of the agencies that are supposed to serve them. That can lead to issues with citizen experience, when legacy technologies don’t measure up to what they’ve come to expect in other aspects of their lives. That’s why it’s important for agencies to modernize the way they do business.

“A pretty good example is the Agriculture Department,” said Wes Jackson, chief technology officer of Three Wire Systems. “USDA had a bunch of paper processes where farmers would bring in these forms they had to fill out, basically the food source surveillance that the USDA has to do to make sure that we can feed a population. And they noticed at one point that the farmers were all bringing in these really nice looking printouts, they all looked the same. And they were hand entering those printouts into the USDA system. And somebody finally thought to ask ‘well, hang on a second; the farmers are all using the same printouts. Are they using some kind of technology we don’t know about?’ And so the USDA had to quickly catch up.”

Jackson said that’s not an uncommon dynamic in the federal government. Many legacy technologies haven’t evolved much since the agencies first invested in mainframes. And it’s not just the technologies themselves; often times the workflow processes are just as antiquated. That’s why he said agencies looking to modernize should consider an intelligent automation framework.

That involves new approaches like artificial intelligence and robotic process automation to automate workflows. But it also means taking a look at the workflows themselves to ensure they still make sense.

“It really starts with understanding your process or your mission, and thinking about the outcomes,” Jackson said. “A lot of the time when people think to go implement automation, sometimes this is the first time in a very long time they’ve actually thought about how does this process actually work. And how does the end result of this process actually drive something important for the organization?”

It’s not unusual, Jackson said, for agencies to realize there’s no reason for some of their processes to be automated, or even to continue. This can not only save agencies time, but also money, because there’s no reason to automate a process when the only justification is “we’ve always done it like that.”

After that, the next step is for agencies to examine their data. It’s impossible to automate without data, and bad data will only guarantee bad results. That’s why 70% of data science is cleaning the data, Jackson said. Part of this is having a consistent collection process. Sources of bad data have to be found and fixed. For example, if your dataset has a column for cities, and the value in one of the cells is “1,” you need to discover where that anomaly originated and why to ensure your data is clean moving forward.

But clean data will never enable artificial intelligence to make value judgements, which is why AI also needs to be designed around human beings.

“It’d be nice if you could just implement an AI solution and walk away, let it run your business. But it just doesn’t work like that,” Jackson said. “For the most part, we like to think that as long as the outcomes are good, then the algorithms must be working. Now that’s a scary trap to fall into. I always say don’t ever accept green for an answer. Just because everything is working, and your dashboards are green, there’s no red on there. If you can’t explain why it’s working today, you won’t be able to explain why it’s not working tomorrow.”

Which is not to say that every decision made by AI needs to be second guessed, especially in the moment. If a self-driving car suddenly applies the brakes, you don’t have to know why right in the moment. But it’s worth going back after the fact to figure out why – was there a pedestrian, or did the visual recognition system get fooled?

Agencies also need to think ahead of time about what new tasks they can apply their workforce to after automating part of their workflow. A lot of employees tend to be hesitant about automation because they’re afraid for their jobs. But automating redundant or repetitive tasks can free these employees up to do more fulfilling work.

“This is another area of missed opportunity, both for individuals and for organizations. Isn’t there something that you wish you could do that would drive more value for your stakeholders?” Jackson asked. “I think that tells us a lot about human nature, and why humans really need automation. Even though they may fear it in the beginning, I don’t think anybody ever longs to go back the other direction.”

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

Related Stories

    (AP Photo/Jose Luis Magana)Congress,

    Flurry of House activity on 2025 federal spending, but not much bipartisanship

    Read more