This week, I heard about one of the rare wins in GenAI. Something that is more than hype. The New York Times wrote about How Airlines are Making Travel Eaiser with AI. While they didn’t get into the technical details, as technologists, we can make educated guesses about how they are doing it.

As product managers, we can learn how to practically solve REAL problems with GenAI. The story talks about four major breakthroughs:

• Holding flights for ~10 minutes to save 13 people from being rebooked

• Prioritizing messages with suggested responses based on urgency

• Making decisions that save fuel

• Using real information to keep people informed, increasing trust and reducing frustration

The teams are hopefully asking:

• How can we reduce the pain of miss flights?

• How can we address urgent issues more quickly?

• Does more information reduce stress when delays to occur?

Why is GenAI better at holding flights than heuristic models that insist on departure times being perfect? GenAI can run simulations faster than existing heuristic models and figure out that holding a flight for 7 minutes might result in 0 rebooking incidents. It has data on: previous flights, current weather conditions, etc. This data can be used to create custom models that run on top of commoditized models.

If you ask ChatGPT how they do it, you get 8 elements:

1. Real-Time Data Analysis

2. Predictive Analytics

3. Operational Research Models

4. Decision Support Systems

5. Communication Systems

6. Buffer Time

7. Priority Rules and Policies

8. Collaborative Decision Making

While there may be some hallucinations in that answer, most of it sense checks. As product managers, we know most technical solutions are “mixed models” that leverage some heuristics and some algorithmic learnings sitting on human knowledge.

The standard of “care” for passengers is changing quickly. This is a win for society and the environment.

News Flash: The airlines started doing this work LONG before ChatGPT hit the market: “ConnectionSaver "identifies departing flights that can be held for connecting customers, while ensuring those who have already boarded the aircraft arrive at their destination on time," the airline said in a press release in 2019 when the first iteration of the software was first released.” - Yahoo! News

Looking at just Chatbots Master.of.Code took a look at all the parts of air travel journeys could be improved by chatbots.

Dive in More:

• GenAI Gazette & Forbes - How United is providing detailed reasons for delays

  • “The integration of GenAI allows the system to scan flight data and generate preliminary alerts, which human storytellers then refine.”

• Simply Flying - Alaska Airlines helps passengers better spend Reward Points

  • “Alaska Airlines is currently testing an AI-powered flight search tool that suggests destinations to users based on their interests, such as adventure parks or even a wine-tasting experience. What’s even more interesting is that it lets users search for flights that can be redeemed from a specific amount of loyalty points.”

• Master of Code - Addresses full lifecycle of opportunities for AI to help travel.

• Connected Aviation Today - Looks at how sensors are added to the mix to improve predictions.