According to OpenAI, Singapore Airlines has doubled the CSAT, or customer satisfaction score, of its Kris chatbot.
The claim comes from a case study OpenAI published on November 10. It describes the airline using OpenAI models and its API (the interface developers use to build products on top of the models) in three areas: customer support, travel discovery and employee productivity. The doubled score is the only number the page offers. It does not say what the score was before, how it is measured, or over what period. The airline's own confirmation is not part of the material.
That matters because a doubled score is a relative figure. Doubling from a poor result and doubling from a good one are different stories, and nothing here says which this is. CSAT also usually comes from short post-chat surveys, which capture only the customers who choose to answer.
Vendor case studies are also selected for flattering results, which is their job. The useful reading is that a large airline is putting OpenAI's models into customer-facing and internal work, not that the improvement is verified.
Airline chatbots have a mixed public record. In 2024, a Canadian tribunal held Air Canada responsible for wrong information its chatbot gave a customer. A satisfaction score measures how people felt after a chat, not whether the answer was right. The more telling number would be accuracy, or how often Kris hands a customer to a human, and neither appears in this account.