ChatGPT: hype or reality for travel?

Don’t believe the hype warns Simon Goddard from Vibe. Any impact is a way off yet.

Whether it is students using it to do their homework or creating works of art comparable to Picasso prepared in a flash, the news is full of stories about ChatGTP or similar AI systems, and their seemingly limitless powers. But what does this mean for travel? And could we start to see the impact soon?

Simon Goddard, CIO of travel search and booking technology provider Vibe, believes that whilst ChatGPT is a significant step forward for artificial intelligence, any major impact could still be a long way off yet.

“I’d urge everyone in travel to try and plan a weekend break or family holiday using ChatGPT. For sure this is much better than anything to date and the potential applications are enormous, but for now don’t believe the hype: what you’re seeing in the news won’t be a reality in travel for a while yet – and to some extent the human element will never be 100% replaced.”  

Aside from a need for the software to evolve further, Vibe – which is UK-based but has clients all around the world – highlights five important factors in particular that mean the roll out of ChatGPT in the travel space will be slower than perhaps some estimate.

Firstly because of a massive barrier to entry in the form of cost that will exclude all but the very biggest of travel companies. Currently only mega-businesses have the R&D budgets and computing power to make true AI possible. For example, Microsoft is set to invest $10 billion in OpenAI as part of a funding round that would value the company at $29 billion.

Secondly because without access to personalised data about the user, ChatGPT (or other AI for that fact) can’t recommend anything actually personalised – meaning that everyone gets offered the same results regardless of age, gender, location or multiple other basic factors that would be necessary to tailor results. In travel this is crucial as most good travel agents know their top clients personally.  For example, they are able to build on previous trip knowledge to recommend places and properties that are immediately suited to the traveller. More advanced factors such as budget,  ages of children, birthday milestones and so on would all add further to planning the perfect holiday.

When it comes to travel, as opposed to picking a birthday present or recommending a restaurant, there’s also still too much risk around getting basic but essential data wrong. For example, imagine if AI gives wrong information on what vaccinations are required for a trip or incorrect advice on visas and immigration rules? Or doesn’t allow enough time for a stopover resulting in missed flights. An additional problem is that currently Chat GPT only knows about information available until 2021 – and more than ever we’re now acutely aware that geo-politics, terrorism, pandemics and disasters can make whole regions no-go areas overnight. This means human checks are necessary and that not only defeats the object of using ChatGPT, it also undermines confidence which in turn reduces uptake until it can have less lag time on knowing about ‘current’ affairs.

Serious concerns also exist that any at-scale provider who offers an affordable version would monetise this by proving biased results to lead people to their adverts or products. ChatGPT is now exploring a subscription model for revenue generation, but it is difficult see regular consumers paying for yet another monthly subscription on top of Netflix and countless others. More likely is that Google or Microsoft Bing will take full advantage of more lucrative advertising models with their offering, especially as these are better aligned with their search engine commercial models. At the moment we understand when adverts are presented to us in search engine results, but these lines could become blurred in the context of an AI response and regulation would be needed.

There are also some potential data privacy challenges presented by this development. AI at this level is going require off-device processing for some time yet. This means that any input, be it text or voice, is going to be sent somewhere to be understood and a response generated. This opens up serious questions about privacy as these communications will contain much more personal information than asking Siri for directions or Alexa to turn on the living room lights.

As a final but perhaps very important barrier to the update of ChatGPT in travel, Simon points to one very big failing: “Right now the tools don’t always lead people to options that are bookable. At this time, ChatGPT only really knows about directing people to big OTAs such Booking.com or Expedia – great for them, but not the smaller players. If the biggies don’t have the product then ChatGPT simply says ‘use a travel agency’ and there’s no deep links to actual bookable options or images. Ultimately we’re all running businesses here, so if we can’t monetise it then what use is this to us?”.

In conclusion Simon sums up by saying: “The future just got brighter for sure and it’s definitely fun to mess around with, but ChatGPT and travel remains just that still: something in the future.”