Artificial Intelligence (AI) and Machine Learning (ML) is a broad domain, and not a new one across the national security and global infrastructure industries. If you ask multiple institutions and individuals what the definition of AI is, their answers will vary.
At its core AI refers to the creation of systems or machines able to solve hard problems and achieve goals – just like biological creatures. Technologies such as ChatGPT, Claude, Gemini, Titan, and other large language models (LLMs) are fundamentally disruptive. Just as the invention of the “horseless carriage” changed the way people used transportation, the release of LLMs is changing the way companies do business and how people live their lives.
In both cases, we need to develop entirely new skillsets to navigate this unique environment. AI changes the way we use data and approach a mission, while also expanding the realm of problems that we can solve. As a result, the techniques successful companies and individuals use to solve complex mission challenges will also change.
What Is Artificial Intelligence?
AI has been called both the villain and the hero since before OpenAI released ChatGPT into the world in November 2022. As a villain, it is charged with taking jobs, being racist in its decision making and responses to users and operating in a non-environmentally sustainable way. As a hero, it is credited with increasing productivity and efficiency in the workplace, being an intelligent sidekick or copilot in our day-to-day activities, and helping users and companies solve the challenges humans have been unable to solve on their own.
The concept of created intelligence is not new. It can be seen in fiction as far back as the Jewish Talmud in the fifth century BCE (where creatures called golems were made of mud and magically instilled with focused ‘intelligence’) and Greek mythology (where the god Hephaestus created Talos out of bronze and injected ichor into him to bring him to life).
AI has existed as a scientific discipline since the 1940’s, when Warren McCulloch and Walter Pitts invented the concept of AI neurons. In the 1950’s, Alan Turing defined his “imitation game”, or Turing Test, as a set of criteria to determine whether a computer demonstrates intelligence by conversing convincingly with humans. In 1957, Frank Rosenblatt built the Perceptron, an electronic device which showed an ability to learn.
Much of the work on AI stalled from the mid 1970’s into the early 1980’s, when interest was piqued again with the availability of microcomputers and silicon chips. AI experienced another pause in investment from the late 1980’s through to the late 1990’s, when the invention of the graphics processing unit (GPU) accelerated data processing capabilities. This pairing with cloud compute offerings and efficient, inexpensive data storage led to a resurgence in interest and investment in AI. As an example of that investment, IBM’s computer system, Deep Blue, used AI to defeat Gary Kasparov in a six-game chess matchup in 1997, and NASA’s Mars rovers Spirit and Opportunity were sent to Mars in 2004 with AI onboard to help them navigate Mars’ challenging terrain without relying on human commanding.
In the push for more generalized AI capabilities, multiple domains have developed, including search and optimization algorithms, probabilistic methods, classifiers and decision systems, artificial neural networks, convolutional neural networks, deep learning systems, computer vision, LLMs, generative AI, natural language processing, and generative pre-trained transformers (GPT). As compute resources have become more available, performant, and inexpensive, LLMs have grown larger and larger, even as the inference capabilities provided by convolutional neural networks have been increasingly optimized in edge devices, where small resource footprints and power consumption requirements are essential.
Who Is Parsons In This domain?
Our team has been developing and delivering AI solutions for our customers in both national security and global infrastructure for over two decades. We combine our deep mission understanding in Intelligence, Surveillance and Reconnaissance (ISR), Cyber, Spectrum Operations, counter Unmanned Air Systems (cUAS), Space, and JADC2 with our AI capabilities to deliver tailored solutions for our customers.
We have proven expertise in six areas of AI technology, including ML, generative AI, computer vision, optimization, natural language processing, and decision support systems. We have delivered operational systems to our Critical Infrastructure and Federal customers, including accredited foundational frameworks and models, which we use to jumpstart customer programs. Our AI Center of Excellence (AI-COE) provides resources to our delivery teams, manages our AI tools, and develops training curricula for our employees to ensure our implementations using AI deliver consistently high quality, secure, and ethical AI solutions. In addition, our employees contribute to and lead within international organizations such as the International Committee for Information Technology Standards (INCITS), where we have four expert employees registered with INCITS AI working collaboratively to develop standards for data protection and the sustainable and ethical use of AI.
How Is Parsons’ Approach Different?
Our approach to AI is multi-faceted. For our customers, we combine our proven frameworks and models with mission knowledge and focused partnerships to craft tailored AI/ML enabled solutions. For our teams, we provide tools such as our AI-COE, our private instantiation of ChatGPT, on demand learning curricula that increases our organization-wide knowledge of AI, and gamification upskilling events to ensure our employees understand the power and peril of AI. This approach ensures that every Parsons employee has access to the tools, training, and support they need to deliver advanced, mission-focused solutions incorporating AI to our customers.
For example, Parsons and cloud partner AWS is currently running an all-Parsons competition using AWS DeepRacer and recently offered another using Amazon PartyRock. These competitions are open to all employees regardless of role or coding / AI knowledge and help participants understand the power and limitations of AI and large language models (LLMs) like ChatGPT. Our “Planet of the Apps” competition encouraged employees to create generative AI-based applications that solved a business challenge, entertained the user, or were just generally fun. A panel of judges evaluated the entries and selected winners for four categories: Most Creative, Most Utility, Most Fun, and Best Overall. In addition, all our employees were invited to play with the submitted apps and vote for the winner of the Fan Favorite award! We had 53 apps submitted and a lively voting competition for the Fan Favorite category. The talented winners, from across the entire company, received bonuses and customized prizes as recognition for their ingenuity.
Why Does All This Matter?
Generative AI is unique within the AI domain in its accessibility to everyone. It augments human capability without replacing it. We recognize the potential in this technology and understand that the technology is only as valuable as the innovative solutions our employees create using it. Our investment in gamification events like the “Planet of the Apps” and DeepRacer contests demonstrate our differentiated commitment to our employees and customers.
By providing our people with exciting opportunities to dive deep into multiple disciplines of AI and prompt engineering, we position them for success by ensuring they know how to apply them. We take pride in our people, their knowledge, and our ability to imagine next in the solutions we craft using technologies like generative AI and ML.