AI is getting stronger in federal civilian agencies

0

In their quest to modernize systems and improve service delivery, federal agency IT and program teams are turning to artificial intelligence (AI) and robotic process automation (RPM). These technologies, different but related, find applications in a variety of situations, find their place in new applications. AI and RPA also find their place in the process of developing new applications themselves.

To gauge the state of AI and RPA adoption by non-defense agencies, Federal News Network…

READ MORE

In their quest to modernize systems and improve service delivery, federal agency IT and program teams are turning to artificial intelligence (AI) and robotic process automation (RPM). These technologies, different but related, find applications in a variety of situations, find their place in new applications. AI and RPA also find their place in the process of developing new applications themselves.

To assess the state of AI and RPA adoption by non-defense agencies, Federal News Network assembled a panel of agency and industry experts and got their point of view. seen.

Jerry Ma, director of emerging technologies at the U.S. Patent and Trademark Office, noted that the scope of prior art examiners to check during the processing of patent applications may have multiplied by thousand since the advent of the Internet. Without technical aids, no examiner can hope to flush out all possible cases of relevant prior art.

“AI, among other things, is what we consider absolutely essential to help examiners,” Ma said.

For Ma, there is no hard line between AI and RPA, and in fact the two types of software can complement each other.

“For example,” Ma said, “you could have an AI rule or an AI model that acts as a trigger to launch an RPA bot. of AI inference work.”

At the Securities and Exchange Commission, officials envision AI and RPA as a complement to people performing mission-related tasks.

“Scaling up is primarily our focus,” said Dr Tanu Luke, the commission’s director of strategy and innovation. She said that AI and ML apply at all levels for the investor protection mission. Luke said she was looking to apply RPA in document-intensive operations, to spare people repetitive tasks like sifting through stacks of paper for particular information.

Luke said where RPA is “prescriptive”, AI is “descriptive”, applicable when “we don’t really know what’s in the data”. Luke cited natural language processing applied to narrative documents as an example of where AI can help the SEC’s review and investigation processes.

Months to minutes

The Government Accountability Office’s chief scientist, Dr Tim Persons, said agencies at all levels are adopting RPA for rules-driven processes with what he called deterministic results. At the GAO itself, he cited a labor calculation that RPA reduced from months to seconds.

For such processes, Persons said, “you know your rules in advance and how to do it. You just need to be a process engineer and code an algorithm.

According to people, the next wave of automation involves larger and more varied data sets, and less deterministic “two plus two equals four” results. He said facial recognition is an example of the second wave as defined by the Defense Advanced Research Projects Agency (DARPA). Applicant screening systems are another example of the second wave. People added that every agency uses wave one or two.

The third wave, he said, is “the stuff of movies as a fully self-contained system of thought.

According to Melissa Long Dolson, vice president of global technology sales for AI Ops and Integration at IBM, another differentiator between RPA and AI is whether the issue of ethics comes into deployment. She noted that AI applications can introduce biases that could produce unacceptable or biased results.

“When you start thinking about AI, you really start to incorporate human intelligence into the logarithms of the models,” Dolson said. “It changes the dynamic because the AI ​​then really becomes a person in itself.” Therefore, the design of the algorithms and the datasets used to train them require careful attention to avoid bias.

“IBM holds many workshops with its customers to ensure that we help them map out the processes they would automate,” she added. “And make sure we’ve built the right logarithms and can train both the AI ​​capabilities and the bots to do it like a human would.”

Set an AI priority

The agencies in our panel apply a rigorous methodology to prioritize their RPA and AI projects in order to obtain the best returns on investment.

Luke said the SEC sees three main buckets. First comes operations and document processing, and their associated repetitive tasks. Second, compliance with executive orders, in particular the application of zero trust principles to SEC networks. Third, various specific mission areas within the agency, such as reviews, enforcement, credit risk management, and liquidity management.

“Our AI applications are more focused on how we can first and foremost enable SEC employees to do the things they need to do faster,” Luke said. The business logic for onboarding new employees or departing employees is on the horizon for AI, she added.

Because the USPTO is primarily funded by patent and trademark application fees, Ma said, the highest priorities for AI and RPA “always focus on what provides maximum value to our stakeholders and This means increasing, he said, not only examiners’ ability to see all possible prior art, but also helping litigation arbitrators make accurate and justifiable decisions.

As noted by Luke, AI and RPA apply to information technology operations themselves, in the case of the SEC to help establish zero trust. People have underscored this point saying, “We see that there is no future in cybersecurity, in running things, that doesn’t involve AI and machine learning.”

Dolson cited another example. She said that across the board, agencies are looking for AI-powered applications to help them determine which workloads are best suited for cloud deployment.

“We help our clients assess the workload. Not only do we provide them with suggestions on what workloads to move and when, but we also help them understand what that job entails,” Dolson said. The algorithms take into account the variables of an application’s infrastructure needs, the data it needs, and even the nature of the source code.

Ultimately, the folks at GAO said that by always keeping the mission at the forefront of thinking, the priorities for what and how to automate and accelerate with AI and RPA will manifest.

Share.

Comments are closed.