top of page

Operationalizing artificial intelligence: Making it work for the mission today

Members of the 153rd Intelligence Squadron perform tasks as part of the space-focused targeting mission at Ebbing Air National Guard Base, Fort Smith, Ark. US Air National Guard photo.

In this Q&A with Ricardo “Rico” Lorenzo, chief technology officer for Parsons, we discuss: how the company is operationalizing AI today; methodology for applying AI to current mission needs; and how it’s connecting AI to legacy systems.

Breaking Defense: What are the threat scenarios today that Parson is helping to address?

<img aria-describedby="caption-attachment-315054" decoding="async" loading="lazy" src="" alt="Rico-Lorenzo (1)" width="265" height="300" srcset=" 265w, 338w" sizes="(max-width: 265px) 100vw, 265px"/>

Ricardo “Rico” Lorenzo is chief technology officer for Parsons.

Lorenzo: With a diverse portfolio spanning federal and infrastructure businesses, we find ourselves addressing a number of different threat scenarios, such as GPS spoofing/jamming, electronic attack, malicious UAS, insider threat, industrial control systems compromise, counterspace operations, cruise and ballistic missiles, hypersonic weapons, cyber-attacks, and biological threats.

To leverage AI in addressing today’s threats effectively, we focus on scenarios with a high “AI- readiness” level. These are scenarios that have the right ingredients for taking advantage of AI today.

Through our years supporting the DoD and IC, we have gained an intimate understanding of our customers’ missions, threats, systems, and data. Even before AI became so mainstream, we were already supporting our customers with the development of solutions for the ingestion, processing, exploitation, fusion, dissemination, and visualization of their mission data.

In many cases we even helped develop the protocols, standards & interfaces for the data they use today. In these cases where we have all the “ingredients”: mission understanding legacy system knowledge & data understanding/access – these are scenarios that have a high “AI-readiness” level.

A good example of this is where we are successfully applying AI – our work around AI-enabled decision support systems. We are actively working with our Space and Army customers on capabilities around predicting friendly and enemy courses of action by analyzing intelligence data. This provides decision-makers with high-confidence courses of action on how best to employ assets depending on where the enemy is likely to act next.

On both of these programs (Space & Army), our intimacy with the mission coupled with an understanding of legacy systems enabled us to quickly apply AI-enabled capabilities to address mission requirements.

Breaking Defense: What are the DoD’s challenges with implementation of AI/ML? Everyone’s talking about the future infrastructure of JADC2 but what can be done with the present infrastructure to create JADC2-like capabilities?

Lorenzo: Amidst the recent focus on AI, there has been a rush to research and develop around the art of the possible. However, we recognize that operationalizing this emerging technology today is a major challenge for the DoD. While billions of dollars have been spent on developing and deploying legacy systems and infrastructure, it is essential to understand how AI can be integrated into these systems without disrupting them.

We define “operational AI” as AI capabilities built to function within real-world hardware and communications constrained environments, interacting with legacy software systems, and capable of operating on disparate and sometimes unreliable sensor feeds while still performing reliably. In order to operationalize AI, one has to understand where and how it will be deployed and used.

The end user is a key consideration as well. The users are not data scientists, nor software engineers. Therefore, we must provide them with appropriate tools and information to successfully leverage the technology in support of their mission. This is essential in building user trust in the technology and making a measurable impact on mission execution.

At Parsons, our role in operational AI is about integrating this critical emerging technology within existing infrastructure. The JADC2 vision relies heavily on established systems undergoing modernization efforts – an opportunity for us to fit AI into the roadmap. For instance, we have been developing an AI enabled EW planning optimization capability in partnership with the S&T and acquisition community to place this capability into the program of record roadmap.

Breaking Defense: Describe Parsons history with AI development and implementation in the defense sector, including infrastructure. What’s it mean to take a “holistic” and “agile” approach to AI?

Lorenzo: I know it may sound cliché, but at Parsons, we have been supporting the DoD with the development of AI capabilities since the early 2000s. Working in partnership with the Services R&D community we were implementing techniques for predictive awareness. As a longtime solutions provider in support of DOD missions across all of the services and the IC, we are often the “boots on the ground” making sure capabilities meet real-world mission needs in the field. So for us, a “holistic” and “agile” approach to AI translates to “how can we quickly adapt this emerging technology to the mission today and in today’s environment.

”There have been tremendous advances and investments in AI across industries over the last several years. We focus on partnering with industry and government customers to quickly adapt these technology investments into their existing operational infrastructure to improve mission efficiency and address capability gaps.

To support this strategy, we have established PALADIN labs – an environment where industry, government, and academia can conduct research, development, and testing in a low-risk venue. Here, they can prototype proofs of concepts using emerging and legacy hardware, AI algorithms, and software technologies with existing government architectures and real data. Through this process, the government can rapidly determine which commercial AI capabilities are worth pursuing.

With facilities in key locations – Aberdeen Proving Grounds, Huntsville, Herndon & Denver – PALADIN labs enable us to develop an AI ecosystem that can be quickly proven against mission systems. As AI implementation becomes more engrained in our customers’ systems and everyday life – especially for DOD but also civilian infrastructure – it is paramount to verify AI models in a low-risk environment before deployment.

Breaking Defense: Final thoughts?

Lorenzo: I think these are exciting times, especially as a technologist. However, it’s essential to be aware of the risks associated with AI implementation and have a framework in place to manage them. Although technology seems to be moving faster than policy, organizations such as the International Committee for Information Technology Standards (INCITS) and NIST are collaborating with the private and public sectors to build standards and tools like the AI Risk Management Framework. These frameworks help the community manage risk effectively.

At Parsons, we recognize the importance of remaining active participants in these forums and leveraging these tools and frameworks to ensure we are responsible and proactive in securing our AI implementations.

I am excited to continue the AI discussion with industry and government customers alike. At Parsons, we are committed to exploring ways to bring this exciting technology to the mission in a responsible and effective manner. We believe that by leveraging our collective expertise and collaborating with our partners, we can develop innovative solutions that meet real-world mission.

Source link #Operationalizing #artificial #intelligence #Making #work #mission #today

4 views0 comments


bottom of page