The great AI travel takeover – should we believe the hype?

There’s a definite ‘Emperor’s New Clothes’ vibe to all this. We’re being encouraged to believe by some that the way we make bookings is going to change forever and if you’re not on board, you risk being seen as a dinosaur, behind the times and perhaps worse - you risk becoming obsolete in this shiny new world.

But now, perhaps more than ever, there is an increasing number of people are starting to wonder if AI’s role in travel is quite the game-changer it’s being made out to be.

Don’t believe the hype (or perhaps take it with a pinch of salt)

To be clear, the evolution of AI has been nothing short of astonishing. Along with the internet, it is undoubtedly one of the most incredible technological advancements I am likely to witness in my lifetime. Yes, even better than when I upgraded my ZX Spectrum for a Sega Master System and discovered I would never have to wait 40 minutes for a game to load ever again.

Yet when I hear sweeping declarations about AI changing how we book travel forever, or the booking process becoming ‘unrecognisable in five years’ time’ as one speaker at this year’s Phocuswright conference proclaimed, a little voice in my head - my inner Luddite, if you will – whispers ‘yeah right – you want to try telling my mum that?’.

AI tech companies are coming under increased scrutiny at the minute, two recent examples including the BBC threatening AI firm Perplexity with legal action over unauthorised content use and Disney and Universal suing Midjourney for copyright infringement. No longer just a byword for innovation, artificial intelligence risks having more negative connotations than positive (many would suggest that is already the case!).

This all suggests to me that the travel industry should navigate the choppy waters of AI adoption with extreme caution. There’s a reason why the likes of Travel Counsellors is thriving – people value the human touch. AI can’t tell you what it feels like to go surfing on the Gold Coast, or experience the sights, tastes and smells of a Hong Kong food market. It can’t give you that authentic, almost tangible personalisation that a human can.

And it seems the outputs are still coming up short, according to a study from Cornell University, which found that OpenAI’s most advanced model achieves only a 10% success rate on complex travel planning benchmarks. Even with all the necessary information, AI failed to create viable plans 90% of the time.

Yet despite the mounting lawsuits, talk of a ‘subprime crisis for AI funding’, a litany of exaggerated promises and the niggling sense that some of this is more smoke than substance, the fervour around AI for many in travel refuses to die down.

So how do we make sense of all this - the inflated claims, the backlash, the slow but steady adoption? They’re not new phenomena. In fact, there’s a well-established model that helps explain exactly this kind of cycle - the Gartner Hype Cycle.

What is the Gartner Hype Cycle?

The Gartner Hype Cycle was developed by US research and advisory firm Gartner in 1995 to represent the peaks and troughs of specific technologies. But it could easily be applied to most things in life –fashion trends, TV shows, boybands, etc.

The five stages of the Hype Cycle

• Technology Trigger - a breakthrough or innovation that sparks excitement early on in its development.
• Peak of Inflated Expectations - where whipped-up media frenzy and hyperbole fuel expectation. Success stories are well publicised whereas failures are overlooked.
• Trough of Disillusionment – interest starts to wane as people realise the tech isn’t as game-changing or exciting as they thought it would be.
• Slope of Enlightenment – people start to accept the tech’s flaws and understand its potential. Second- and third-generation versions may start to appear from technology providers, ironing out previous issues. Some companies start to invest in the tech while others remain cautious.
• Plateau of Productivity - adoption of the tech hits the mainstream and becomes commonplace.

How AI aligns with the Hype Cycle

Since late 2022, AI’s journey has mirrored the Hype Cycle almost step for step. The dawn of natural language models and machine learning spawned an outpouring of articles, podcasts and blogs which gradually filtered into mainstream conversation. Travel firms got in on the act too, with the likes of Kayak and Expedia being early adopters. AI was going to reinvent how holidays were searched, planned and booked - as neat and tidy as that. Tools like ChatGPT and Google’s AI Overviews would bypass websites entirely, understand a user’s intent, suggest hotels, compare flights and somehow do it all with perfect personalisation and zero hitches.

But it wasn’t all plain-sailing – far from it.

As I noted in this blog about AI trip planning a couple of years ago, chatbots were prone to some almighty gaffs. Here’s one of them:

Google rolled out Bard (now Gemini), then promptly rolled it back in again when the glitches started piling up. Bing Chat (now Copilot) didn’t fare much better. It became obvious that while the technology looked promising, it wasn’t yet ready for full scale adoption across all industries. Cue the Trough of Disillusionment phase.

That said, this was happening in 2023 – and two years are like ten in tech terms. It’s fair to say there are far fewer glitches nowadays (well, Mr Musk is still an outlier) and the technology has undoubtedly come on leaps and bounds. Natural language models have improved. Machine learning systems began learning from user behaviour in more meaningful ways. Take away the grandiose proclamations made at tech events and AI found itself following the Slope of Enlightenment: the point in the cycle where lofty promises make way for real-world integration.

Plateau of Productivity or just the beginning?

Where you feel we are now on the Hype Cycle is going to be dependent on the changes that you've made in your own travel company. We know of a few businesses that are using AI as the foundation for almost everything they do. Some have made decent progress but aren’t seeing the results they were hoping for, while others are yet to make a move and are choosing to wait a little longer. These examples tend to reflect more obvious investments in AI, aimed at transforming a particular process or system.

Generally, I think we have reached the Plateau of Productivity stage. Most brands are using AI to some extent, be that booking systems, marketing content or customer service support. Consumers, and even some travel businesses, may not even realise it’s there, and that’s the point. The technology works best when it’s invisible - a ‘behind the scenes’ entity, allowing travel agents and account managers to focus on building relationships with their customer base.

While the tech is more capable and reliable than it was even a year ago, there’s a sense that things are no longer moving at breakneck speed. In fact, some experts suggest we may have hit a plateau in terms of model development itself.

But the buzz at Phocuswright Europe felt very much like a return to the Peak of Inflated Expectations – a belief that something even bigger is just around the corner. Understandable, especially if you're in the business of building the next big thing.

I keep thinking about that word ‘unrecognisable’. Am I that much of a naysayer/luddite to think such a word is overly hyperbolic? Was I perhaps one of those people who thought: ‘this world wide web wotsit? Nah mate – it’ll never catch on’.

I don’t think my cynicism stems from a lack of forward-thinking, but rather from what I see around me, personal experience and what I’m reading on the world wide web wotsit – and that is that a) people of all demographics still have plenty of misgivings about using AI for bookings and b) travel agencies are making a comeback. As for booking a holiday without ever visiting an actual website (which is what could happen, especially if Google decided become a full-blown OTA), that might work for a very simple, single-destination trip, but I can never see it becoming ‘the norm’ for family holidays or multi-destination trips – at least, not in the next ten years or so. But, hey, if anyone is reading this in 2035, give me a shout and let me know if I was wrong.

So what will the future really look like?

AI is not going anywhere. Nowadays it’s pretty much part of the furniture. But travel, in my opinion, will always remain a people business, certainly when it comes to the big trips where more is at stake. Because when it comes to travel, the safest pair of hands will always be human.