AI in the Control Tower: A Framework for Integrating AI to Serve Humanity
Peek Counsel is an independent consulting firm that guides organizations during their transformational processes to become equitable and sustainable enterprises of the future.
Have you noticed an increase in news stories about plane crashes, malfunctions, or near-collisions? I have. I’ve seen reports ranging from social media posts to articles in reputable journals. At first, I wasn’t sure whether this surge in coverage was simply the result of algorithms distorting reality or if it was a legitimate trend. So, like many people do these days, I turned to Google. The AI-generated summary at the top informed me that plane crashes have actually decreased and that air travel remains relatively safe.
However, I scrolled past the summary to explore more sources.1 One article by PBS, titled "What’s Behind the Alarming Rise in Near-Collisions of Commercial Airlines?" caught my attention. It transcribes an interview between Miles O’Brien and reporter Geoff Bennett, discussing the increase in close-call commercial plane collisions. The article also references an investigation by the New York Times, which revealed that in August 2023, there were 46 near collisions in July alone. The piece includes animations showing how close these potential accidents were to occurring. According to the Federal Aviation Administration (FAA), "close calls involving commercial airlines have been happening, on average, multiple times a week" in 2023.
So, what’s causing this? The main factor is human error, particularly from overstretched air traffic controllers (ATCs). It’s no secret that air travel has surged post-COVID. The airline industry is enjoying record profits, with North America alone contributing $14.8 billion in revenue. While airlines are seeing huge financial gains, this growth is also leading to significant challenges, such as shortages of both pilots and ATCs. These shortages stem from early retirements and the lengthy training process—becoming a fully qualified ATC can take four to five years. The wave of retirements, combined with a lack of new hires, has forced pilots and ATCs to work longer hours, which is impacting safety.
Where AI Comes In
Studies show that 80% of all aviation accidents are due to human error. While this can increase when other factors are not ideal (i.e. overworked staff), even in low incidents a human mistake is always the main culprit. In addition, the PBS interview mentions that the reputation of the aviation industry as a dinosaur proceeds them, as they still rely on 1950s technology which increases the likelihood of errors. The situation appears to be a perfect application for the ever-evolving artificial intelligence (AI), especially as the International Air Transport Association (IATA) projects that flying will double in the next 20 years.
AI had its inception in the 1950s, but it has only recently begun to be integrated into the aviation industry. Currently, AI automates aircraft maintenance and ground handling, with the use of drones in some cases. Airlines are also beginning to integrate AI into their operations such as Alaska Airlines use of Flyways AI2 and Delta Airlines use of Starlink as the new in-service wi-fi provider. Some Airlines are even using AI to train the new workforce. What if it had another application?
Airlines could automate air traffic control. Radios are still used between pilot and ATC, and a driving factor of miscommunication that leads to accidents. AI should be creating an elevated system that works for pilots and ATCs by improving efficiency and safety, and lowering miscommunication incidents. Automated systems for mass transportation already exists, such as the technology that runs Dubai Metro. Similar technology can be configured to air traffic control. The use of AI to fill in ATC roles would compensate for the expected deficit in aviation staff relative to flight demand. According to the Aviation Talent Forecast, by 2024, the aviation industry will need 1.3 million civil aviation professionals in the next decade. If these roles cannot be filled in time by humans, automation may be necessary.
While the automation of some tasks will result in job displacement and job loss, it does not mean workers are disposed of. Training programs could create new types of jobs and reskill displaced staff, enabling them to transition into new roles that complement AI technologies rather than compete with them.
However, a key consideration in implementing AI is addressing the accompanying security risks. As AI technology becomes more interconnected, the potential for cyberattacks increases. Strong legislation and policies at all levels of government are essential to safeguarding against these threats. The recent shift in administration has seen a reversal of some AI restrictions set by the Biden administration, but certain policies remain consistent, such as federal support for expanding data centers needed to support AI development. This could benefit the aviation industry, increasing the likelihood of federal support for AI integration into additional processes, like air traffic control.
Sirius Intelligence (SI) AI Framework
Considering the complexities involved in determining when to apply AI, the aviation industry serves as a valuable case study for developing a decision-making framework to guide the adoption of AI technologies. The process of Peek Counsel’s Sirius Intelligence (SI) framework for AI application is illustrated in Figure 1. The framework is meant to guide industry and businesses who are in limbo when it comes to AI.
Figure 1. SI’s AI Application Framework
Based on this framework, industry and businesses considering when and where to implement AI technologies can base their decisions on three key factors:
Workforce Shortage
Assess the availability of skilled workers and the need for automation to address gaps.Audit of Automatable Positions
Identify positions that can be automated and evaluate the impact on overall operations.
Skillset Shortage
Analyze the gap in required skills and the time needed to upskill current employees or train new ones.
Outdated or Inefficient Systems and Processes
Determine whether existing systems and processes need upgrading or streamlining through AI integration, based on efficiency and safety.
Cost/Benefit Analysis
Conduct a comprehensive analysis that includes the potential for increased cybersecurity needs and the establishment of a policy tracker department.
If all the following conditions are met—1) workforce shortage, 2) skillset shortage, and 3) outdated systems impacting efficiency and safety—then it’s a strong case for gradually introducing AI across all relevant areas. If only the workforce shortage condition is met, assess which roles can be automated and compare the costs and time involved in automating these roles versus hiring and training new staff. If the skillset shortage condition is met, calculate how many months it would take to upskill current employees or train new staff. If a significant staff deficit is expected to last more than 30 months (or as per company policy), then consider applying AI to address existing gaps, while gradually upskilling staff to complement AI rather than compete with it. Finally, if the outdated systems condition is met, perform a cost-benefit analysis on implementing AI to upgrade software across operations (e.g., payroll). This analysis should include considerations for enhancing cybersecurity and establishing or strengthening (hiring) a policy department.
An essential component of this framework is for companies to establish their own baselines and thresholds to assess the findings. For example, a company's policy on maintenance staff levels will be crucial in determining how long it can effectively operate with a staffing deficit. Similarly, whether or not a company outsources certain tasks will effect the cost/benefit analysis in establishing or strengthening a policy team/department.
If AI is to shape the future, let it complement and enhance humanity rather than pose a threat. Stay tuned for more blueprints and frameworks to help companies successfully navigate the new era.
Since I phrased my question to AI to search for trends in place crashes, the summary omitted information on the increase in near-collisions. I still find it peculiar that the AI summary was not more nuanced.
Flyways AI platform is making flights more energy efficient with reduction in emissions and better mapping routes.