3 minute read
Soaring higher: How AI and LLMs are transforming the airline industry for Crayon customers
Words by Armin Haller, Head of Centre of Excellence for Data and AI at Crayon APAC
Artificial intelligence, and in particular large language models (LLMs), have significant potential to revolutionise the airline industry. Whilst AI and neural networks, the models underlying LLMs, have been in existence for many years, it is in 2023 that they have reached mainstream prominence with the introduction of ChatGPT. This AI model, built upon LLM technology, has surprised even experts with its remarkable accuracy and, in certain scenarios, demonstrates capabilities close to what experts call general artificial intelligence.
The availability of LLMs in the cloud, such as those offered through Azure OpenAI, is bringing about a transformational shift. These models have the capacity to process and understand vast amounts of language-based data, enabling any organisation, and in particular those dealing with many customers, such as airlines, to benefit from unprecedented productivity gains.
This period of accelerated productivity that we will witness in the next few years through AI can, in some way, be compared in impact to the industrial revolution. By harnessing LLMs, the airline industry can leverage AI capabilities for various applications. In fact, let’s ask a Large Language Model what it can do for the airline industry.
“First, it provides customer service and support. LLM-powered chatbots and virtual assistants provide 24/7 customer support, handle inquiries, offer personalised recommendations, and deliver realtime flight information. Second, it allows sentiment analysis and social listening. LLMs analyse customer feedback and sentiment from social media, reviews, and surveys to understand preferences, monitor brand sentiment, and inform decision-making. Next, it enables personalised marketing and recommendations. LLMs process customer data to enable personalised marketing, targeted offers, promotions, and tailored travel recommendations, driving customer engagement and loyalty.
Then, it can generate operational efficiency and predictive maintenance. LLMs optimize operationsbyanalysingmaintenance logs, sensor data, weather forecasts, and flight schedules to predict and prevent disruptions, enhance crew scheduling, and improve maintenance processes. It also offers demand forecasting and pricing optimisation. LLMs forecast demand accurately, helping airlines optimise pricing strategies, dynamic fare adjustments, and revenue management for maximising profitability. Finally, imagine natural language interfaces and voice assistants: LLM-powered natural language interfaces and voice assistants enable seamless interaction for passengers, facilitating flight searches, booking management, real-time updates, and access to travel information. Indeed, AI can provide significant ROI, improve customer safety or customer satisfaction, for all of these use cases.”
While any use case that involves text and spoken words is now best solved by LLMs, not all of these aforementioned use cases are necessarily best served by a Large Language Model. Demand forecasting, for example, can be solved best by regression models and other Deep Learning Models rather than LLMs, and operational efficiency and predictive maintenance is best solved through computer vision models. Crayon will help you to determine how a given problem can be solved with AI and which AI model best solves a use case.
For example, Crayon successfully collaborated with a major international airline on a project aimed at addressing the challenges associated with manually monitoring numerous video feeds that record the activities performed around an airplane at the gate. We developed an advanced AI solution that leverages computer vision and machine learning algorithms to effectively monitor security cameras’ video feeds and autonomously oversees various critical aspects, such as identifying catering supplies or specific aircraft-related objects, detecting unauthorized access or suspicious behaviours, tracking cabin crew status and location, monitoring aircraft door status to ensure it is not left unattended, and distinguishing different operational staff while maintaining a comprehensive register of entries and exits.
Our solution greatly assists the security team and on-site experts in making informed decisions, reducing the amount of time and staff required for surveillance without compromising performance. Additionally, the implementation of AI technology significantly enhances accuracy and objectivity in the monitoring process.
The Contributor
Armin Haller is Head of Centre of Excellence for Data and AI at Crayon APAC. He is currently building up a team of data scientists, engineers and solution architects to solve challenging problems using AI and excellent data management. Armin describes himself as an ‘evangelist’ of using technology to improve knowledge sharing between humans. Apart from publishing articles in peer-reviewed conferences and journals, he has been passionately supporting the Australian Government in defining strategies for the use and the publishing of Open Data since 2012.