How ChatGPT can help modernize defense & aerospace
StoryJune 14, 2023
Ian Ferguson
Lynx Software Technologies
ChatGPT – an artificial intelligence (AI) chatbot – has emerged as a potential game-changer in the adoption of AI across all industries. For aerospace and defense, it can assist in increasing automation and data-driven analysis from the supply chain to aftermarket services.
In today’s rapidly evolving digital landscape, artificial intelligence (AI) is changing the way businesses in all industries operate. Aerospace and defense (A&D) is no different: AI solutions can increase operational efficiency, enhance safety, and amplify human teams.
Chatbots, AI-enhanced computer programs that simulate real-time conversation with humans, have emerged as powerful tools that have already been adopted in a wide array of industries, from ecommerce to healthcare to hospitality. Chatbots leverage natural language processing (NLP) and machine learning (ML) algorithms to understand user input and provide relevant responses.
ChatGPT, arguably the best-known chatbot today, is a versatile tool for many industries looking to automate tasks and enhance productivity. Developed by AI research lab OpenAI, ChatGPT is a large language model based on deep-learning technology and trained on vast amounts of data, infusing it with the ability to understand the context of a conversation or other interaction. Its core API is available for developers interested in incorporating this functionality into their own proprietary systems.
But is there anything radically new with ChatGPT? Not really, it could be argued. It does use a model (known as “transformer”) at its core, which means it analyzes information at the sentence level as opposed to individual words. But the main differentiators is the massive data set of information it pulls from and the scale of the user base (which will, in turn, create even larger data sets of information).
ChatGPT in A&D
When thinking about AI applications for the A&D industry, automated robots and autonomous aircraft are more likely to come to mind than chatbots. However, where robots and self-navigating jets might still be years in development, chatbots are assisting engineers and others in the industry today.
This industry, like others including the medical and automotive sectors, requires strong certification. At the center of this requirement is proving that system requirements match completely to the implementation’s behavior. The nature of AI/ML models like the one at the core of ChatGPT is that it is impossible to reverse-engineer the process to determine why the output was produced a particular way. Overcoming this barrier will be the key to how broadly the technology can be adopted into the mission-critical systems that keep humans safe and secure. So, it is unrealistic to expect an immediate removal of a human pilot (either in the plane itself or flying the craft from a remote position) anytime soon.
The consulting firm McKinsey found that more than half of A&D companies it surveyed on the subject have at least started to use AI, deploying it across R&D (71%), supply chain (65%), manufacturing (61%), support functions (57%), and aftermarket services (51%). Let’s dive deeper into how a chatbot like ChatGPT could be applied in these areas.
- Research & development: In the design and planning process, chatbots can streamline a number of functions, including concept ideation, material selection, and design validation. For example, engineers could use ChatGPT to generate new ideas for airplane features by providing it with the design specifications or performance goals, which the chatbot would then use to provide information on existing features or new ideas to incorporate into the design.
- Supply chain: Chatbots can streamline a manufacturer’s supply chain by providing real-time updates on inventory levels and production or delivery schedules. This information can then be shared with suppliers, other manufacturers, and distributors, enabling them to make informed decisions and avoid delays. Chatbots can also use predictive analytics to forecast supply and demand, enabling customers to maintain the correct inventory levels and optimize production schedules. This use of AI can help avoid running out of a specific part (or sitting on overstock), reduce costs, and improve efficiency.
- Manufacturing: On the factory floor, ChatGPT can help detect defective components by analyzing data from various sources such as sensors, visual inspections, or testing data. It can alert quality-control personnel to any potential issues early in production to avoid costly errors down the line. Additionally, the technology can assist in identifying the root cause of defects by analyzing data and identifying patterns, helping quality-control teams understand what’s behind the defect and decide on appropriate corrective action.
- Support functions: Streamlining administrative-support functions such as HR [human resources] and finance is an ongoing process for any business, and A&D companies are starting to take a page from other industries and examine digital ways to increase efficiency. ChatGPT can be used by management to analyze employee-retention trends to identify those at greatest risk of leaving, by procurement to automate parts of the bidding and proposal process, or by finance to analyze workflow for better cost management.
- Aftermarket services: The McKinsey report found that the services groups remain laggards in terms of their adoption of AI. Even so, ChatGPT and other AI-enhanced platforms can be used to analyze data to better understand which services lead to greater customer lifetime value, to streamline customer support, or to develop and manage a blockchain to create transparency across the entire product life cycle, from supply chain to service records. (Figure 1.)
.jpg)
[Figure 1 ǀ Defense and aerospace organizations can leverage AI-enhanced platforms to analyze and help manage data at all levels.]
It is important to note that ChatGPT is just at the start of its journey. The nature of these systems is that as the model gets refined or trained, the answers will continue to improve. That being said, let’s think more broadly about what might be possible.
We are also seeing ChatGPT create software applications. Could some version of this tool help convert legacy code into newer, better-supported languages? The nature of our industry is that code is expected to run for decades. People move on or retire. Could there be an opportunity to help move Ada code into a language like Rust?
Aircraft and unmanned aerial systems contain hundreds, if not thousands, of sensors, cameras, radar, and other systems that generate vast amounts of data that can be used to detect potential threats. These threats can range from a potential system failure to a network vulnerability. Systems can learn what “normal” looks like and then identify unusual data patterns and abnormal behaviors that might indicate a security breach or threat.
To accomplish this task, the technology could be trained to monitor network traffic to detect potential cyberattacks. Unusual network activity, such as spikes in traffic or unexpected patterns in data transmission, could indicate a looming threat, and ChatGPT could be trained to alert a human security engineer to investigate further.
In addition to monitoring an organization’s network and data traffic, ChatGPT can also be used to scan external sources of information, take note of global cybersecurity incidents and patterns, and identify potential sources of threats, such as suspicious IP addresses. This data can then be fed back into a predictive-analytics algorithm to recognize potential vulnerability to emerging threats.
One of the gates determining how broadly AI and ML will be deployed in systems is how/if certification processes adapt. We have seen proposals in Europe that provide guidelines as to how to embrace AI and ML as a powerful assistant to humans. This process will evolve, and we will see a time when AI and ML are at the core of mission systems in the air. Unlike the automotive market, which has struggled to achieve its vision of fully autonomous travel, it is arguably easier in the air, primarily because there are fewer unpredictable items (like humans stepping off the curb) to worry about. For this to become a reality, however, the technology will need to evolve to a point where the AI/ML decision-making is bounded – both in time and in terms of impact on other systems – and the system can be proved to be fit for purpose.
By all accounts, the A&D industry remains in the early exploration stage of implementing AI. But the sector has always been one of innovation. While perhaps the industry isn’t at the forefront of adoption, A&D stands to reap significant value from ChatGPT and other forms of artificial intelligence. Industry reliance on paper-based systems, siloed operational platforms, and other traditionally manual processes will be challenging to change. However, new generations of engineers and approachable AI platforms like ChatGPT may smooth the runway toward more widespread adoption.
Ian Ferguson is vice president of marketing at Lynx Software Technologies, responsible for all aspects of the outward-facing presence of the company to its customer, partner, press, and analyst communities. Ian also is responsible for nurturing Lynx’s partnership program to accelerate the company’s engagement in mission-critical systems. Readers can contact him at [email protected].
Lynx Software Technologies • https://www.lynx.com/
