Although the term has been lingering in academic and science fiction settings for over 70 years, it seems to me everybody realized two days ago that AI is the new magical panacea that will end profitability issues overnight, replacing “big data” (which now sounds so 2016). As such, everybody and their granny is adding into conversations, with utmost nonchalance or even an expert stand, the trite two-vowel expression… sometimes without a clue of its true meaning, others in a completely disconnected context. Even worse, there’s a legion of unethical hustlers selling any pathetic automation as “AI-powered”. This trend grew during the past decade or so, due to the proliferation of mass social media, of course. Who knows what the buzzword will be next year!
As much a science fiction lover as I am, my training as a (failed) engineer denies me the luxury of contemplating AI (or any other technology) under a romantic light. The practical approach is all that matters in business, and we’re far (in the parsec magnitude) from seeing a computer brain cheat us into believing we’re actually talking to another person (passing Turing’s test, in short). No Skynet, no Voight-Kampff machine anytime soon, folks. So I am shocked to ascertain how much we’re behind in AI’s application all over the travel industry.
Different flavours of intelligence
Much like human intelligence is no longer associated only to logical proficiency, in real life AI is not a single item that powers robots’ minds. The term entails several areas and techniques such as neural networks, machine learning, deep learning, image recognition, and so forth. Each of them has specific uses, and so far it’s (almost) impossible to combine two or more in a single application at the same time. As a result, the different flavours of AI can focus in one -and one only- task. Of those flavours, I’d like to point out the two that are having (or will have) deeper impact in the industry’s bottom line. One is machine learning, a series of algorithms and programming methods that allow computers to detect patterns in data, learn from them, and take a decision or predict an outcome. The other one is deep learning (actually a subset of ML), which works under the hood of recommendation systems like you’ve seen in Netflix or chatbots, for instance. Oversimplifying it, DL addresses data modelling in a “different, better” way than ML: it mimes human brain’s structure, and it’s capable of much harder tasks than ML, like governing self-driving cars or simulating (rather badly, so far) a conversation.
If you’re still wondering what’s the relevance of AI in our travel world, see the graph below, from a study >> published in April 2018 by the McKinsey Global Institute. The word “prediction” should have alert you, but check this out:
This is an estimate of the global impact of AI by industry in 2020. Can you see the implications in marketing, customer service, pricing and operations? Have you noticed the figures in the top blue square? Allow me to reprint them here: $0.6 – $0.8 TRILLION. That’s the value estimated for companies already using some form of AI.
Are travel enterprises monetizing AI? Airlines and large hotel chains certainly are, but the rest of the bunch is not following suit. Let me give you an example: in April 2018 I assisted to a conference in which Mr Peter Mansour, recently appointed Director of Product management at Hotelbeds, presented their current state of affairs in relation to Big Data and AI. I was surprised to learn that they just started working on these things -filling inventory gap, in particular-, as a data science team was ferried from GTA (with a year’s experience under their collective belt).
Sure enough, there are faster, leaner, agile -or simply bigger- companies that convert AI into currency, just like Almundo.com hints in their discreet postings on social media, or in their job ads. But the vast majority of tour operators, bedbanks and even OTAs are not profiting from AI as they could (should!). Which is astonishing, given the money-landfall potential. I found an interesting report by Mr. Scott Nyquist from McKinsey (see it here >>) that includes the following graphic:
Amazing, right? Now, to put you in context, Deep Learning is way out of our league, but here at REVVA we’ve been experimenting with ML algorithms since mid-2017. In particular, we tested prediction models towards buying habits, as well as sales/demand forecasts, which are -in a much smaller scale, of course- more advanced than Hotelbed’s example. It’s not gratuitous bragging… Well, it is to some extent, but what I meant to convey is this: what was previously expensive and complex, now it’s affordable even for SMBs.
Replicant cleverness against human obstinacy
Algorithms, machinery and infrastructure comes with a smaller price tag every day, so how come AI is not yet largely implemented in the travel arena? Is it lack of staff?
I have a theory, structured around three hypotheses:
1 – There’s no AI without an analytics base: if an organization lacks a data-driven culture (see my article on the subject >>), if there is no decision-making based on data, there’s no point in going straight to the AI solution from starters. It would be like sitting in a F1 car a kid unable to ride a scooter. After all, AI is about making decisions, selecting A or B: same as analytics, but quite a bit more complicated (and effective).
2 – There cannot be analytics without data collection: as much as a human brain needs sugar to run, AI needs data to feed and learn from. Large quantities of it, actually. If an organization does not gather and process adequately its own generated information, there’s nothing to analyze. Quite a moronic statement, I realize. How come even large bedbanks and hotel chains are not doing it, then? I suppose it comes down to hypothesis number three.
3 – To collect and process data, as well as make use of it, you need PEOPLE. The right people, mind. People at the top echelon who finally realizes the need of data-driven decisions; people who designs and executes the data project, people that carries on with data tasks.
As you realized, it´s a vicious or virtue circle, depending on the outcome. Funny thing is, small to medium companies need not go over 1 and 2; they simply can start profiting from AI by adopting solutions like REVVA. Alas, such companies suffer even more the third hypothesis, as they are less prone to risk for obvious reasons.
Hence, my theory is simply this:
[ctt template=”3″ link=”6c1US” via=”yes” ]AI is not widely implemented in the travel industry yet because people is -still!- reticent to changes.[/ctt]
From the late nineties, there have been so many major disruptions in the travel industry that you’d think people already adapted to such instable environment. Not so. Quite the opposite: there were so many scams and futile upsells that managers, CEOs and founders are understandably disinclined to invest in new stuff. I can relate to that, but they’d better do proper research if my word is not enough. They will come to the same conclusion:
[ctt template=”3″ link=”hfcU5″ via=”yes” ]Analytics and its daughter AI are here to stay in the Travel Industry, like it or not, believe it or not.[/ctt]
No need to rush, though. In its current state, AI cannot disrupt anything. There’s no revolution going on, no robot is going to replace you anytime soon, subscribing to a chatbot service won’t impress your clients to the point of impulsively buy from you. But it would be silly to ignore the competitive advantage data can bring to your plate, especially in a kill-or-die colosseum like our modern travel trade.
I know, it’s all new and difficult to comprehend, the investments would be significant, the effort considerable. But you must realize that any amount of data that is expensive to collect and store, is by its very nature valuable and profitable. It’s an asset! Next, putting the asset to work, with analytical automations, would be an ideal setting. Finally, before your competitors eat you off (not necessarily global OTAs, also smaller fish brandishing weapons like REVVA), if you’re toying with the idea of benefiting from AI, I urge you to take your first data-driven decision. Review the figures from the graphics above, read a few more articles about the matter, try to establish a cost/benefit relationship of adopting analytics solutions (or pret-a-porter systems like REVVA), then ACT. Your first step should be to create your data infrastructure, if you didn’t do it already.
Otherwise, yours will be yet another organization where human stubbornness overwhelms AI. Time will tell who the winner is, but I know where to place my bet.
Thanks for reading and sharing!
Marcello Bresin