Building a sustainable AI system for Travel Product Distribution is going to be on a lot of people’s minds in this coming year and beyond. So, what do we need to get right if we are to be players?
This post intends to outline what is required and, from a management point of view, what needs to be understood and deployed to make AI work and be truly valuable.
I have previously opined that I do not like the term AI meaning “Artificial (machine) Intelligence” (AmI). I prefer to use the term “Assisted (human) Intelligence” (AhI). Much of what I will add will be in the consideration of the humans in the overall process.
When considering AI, generically it becomes clear that a pure AmI is impossible today.
Any digital system that has AhI pretensions for travel distribution should have the ability to integrate a data supply chain system to effectively gather, process, and utilize data from multiple sources. The system should be able to handle large amounts of data, both structured and unstructured, and integrate it in real-time. However, we know that Travel has three notorious complications:
- Lack of sharing of data across the industry. Thus, large scale data sets are hard to come by
- Silos of data. Thus, it is hard to correlate data in any concerted manner.
- Gatekeepers of distribution. These range from the GDSs to Google.
A good data supply chain system should be able to interface with different systems internal and external such as booking systems, operational systems, and revenue management systems to gather data.
Image generated from OpenAI Labs DALL-E with the instruction: A beautiful illustration of Artificial Intelligence in airline distribution. https://openai.com/dall-e-2/ Text checked predicted to be 99.98% Real. https://openai-openai-detector.hf.space/
Any player must have at their disposal a technology platform to process this data using techniques such as machine learning, and predictive analytics. These insights should be used to optimize pricing, inventory, and operational uses. The key elements of such a system must address the following:
- Data Quality: The quality of the data provided by the service provider is of paramount importance. This includes factors such as completeness, accuracy, and consistency. There is considerable controversy in this area. Who are those players who are responsible for data quality? One such company under the microscope is www.scale.ai for their use of testers and data teams in places like Venezuela.
- Data Security: As we have seen all too frequently in travel, massive data breaches can occur. Thus, any system should also have necessary security measures to protect the data and its ability to be compliant with regulations such as GDPR and CCPA as well as regulatory oversight and reporting that has become more and more important.
- Scalability and Flexibility: Your data needs may change over time, so it’s important to ensure that the system can scale up or down as needed. Legacy systems are going to struggle – perhaps this is the push to get the last of these systems to the graveyard. The system must be able to adapt to changing project requirements.
- Integration: Other in-house digital systems such as CRM and marketing automation systems need to be integrated.
- Support: A dedicated team to support the business and resolve any issues in a timely manner.
But what of the human involvement? In my view humans play a critical role in the success of AhI in travel distribution in several ways:
- Data preparation: AmI algorithms require large amounts of data to be trained and to function effectively. Humans are necessary to prepare and curate this data, which includes cleaning and organizing it, so that it is suitable for use in AI models. In our consulting work at T2Impact – this has been a constant challenge.
- Feature engineering: Humans are necessary to identify relevant data features, extract them and transform them in a way that can be useful for the AI system. This is important but it is my belief that the data needs to ultimately talk back to you. This is a 2-way street.
- Model building/validation: Humans are responsible for building and testing AI models, as well as interpreting and validating their results. This requires human expertise in machine learning and statistical modeling, as well as specific knowledge of the relevant industry sector. This is a skill that is still very rare. There are few players who have this capability.
- Integration and deployment: Humans are necessary to supervise/integrate machine models into existing systems and processes, and to deploy them safely and efficiently.
- Explainability and interpretability: A good system must provide predictions and recommendations but for them to be used, humans must understand the reasoning behind them. Humans are necessary to explain the logic and decision-making process of the system and to ensure its outcomes are aligned with the business objectives.
- Compliance and ethics: Humans are responsible for ensuring that all digital systems comply with legal and ethical standards, and that they don’t lead to discrimination or other harms.
- Human oversight: Even the most advanced systems are not (yet) able to replicate human intuition and creativity, hence the need for human oversight to ensure the decisions and actions of such a digital system are aligned with business objectives and also consider its impact on customers, both internal and external.
In summary, an AhI system for travel product distribution must have the ability to integrate a data supply chain system that can gather data from multiple sources, process it using advanced techniques, and provide insights to optimize distribution including personalized recommendations and offers to customers. All aligned with management and human needs both internal and external.
Perhaps the best way to think of AI is that it is just a tool. AhI can automate and optimize many tasks in travel distribution and thus improve the quality of the travel product – and the business providing it – and make the customer experience truly great.
