What is more natural than a conversation? This may be the biggest promise of generative or autonomous AI.
Omri Morgenshtern, CEO of Agoda, gets to grips with the possibilities of Generative AI in travel in the short to medium term.
It seems there are two distinct camps when it comes to the future of Generative AI – skeptic and optimist. It’s true that in the long-term, no one knows what it will mean for humans, jobs, and industries, and that it is probably going to be the biggest tech leap of our generation. Many conversations about the future of AI talk about job losses and malicious tech. In travel we hear projections about hotels being run almost completely by software with human-like interfaces to book and explore. Maybe that is possible, but it’s quite far-off.
I think the interesting question now is what it means in the short- to medium-term. Say in the next one to five years. This is the critical period for any business looking to stay ahead of the curve. Travel has always been an industry in which doing something a fraction of a percentage better than the competition can crown you the winner, and these tailwinds can be scaled very quickly to huge wins.
I see a few areas that will be disrupted in this medium-term period:
The rise of the automated developer
More than anything, for me, ChatGPT is a very impressive proof of concept of how good a machine can become in things like writing code and fixing bugs. Tools like GitHub Copilot and others are stretching the limits in the medium term. If you can describe in words carefully what you want to happen (call it good old product work), the code will write itself. Just to be clear, we are far from there yet, and likely will not be 100% or even 60% there in the next one to five years. But I think that reaching 50% automation – or doubling the developer efficiency – is more than doable. Some companies might outsource this, and others do it themselves, like we try to at Agoda.
One thing is clear to me: In the next few years, we can increase our development velocity tremendously. And faster velocity means more experimentation, more features, a better understanding of the consumer, and ultimately improves our ability to sell to and satisfy customers.
Automation of content generation
In travel, this might be hotel descriptions, itinerary planning, reviews summarizing or writing aids. All of those are now much faster to generate, and the speed and average quality will only increase further. Most of the big players already have some capabilities in auto-generated content, and AI will enhance that. But the main beneficiaries will be the smaller players and individual entrepreneurs who will be able to do more with less resources.
Conversational interfaces
What is more natural than a conversation? This may be the biggest promise of generative or autonomous AI. Some people are still using travel agents, partially because travel agents allow us to talk through our travel plans in real time. Will AI make those who are hesitant to book online more comfortable?
The first step is to get consumers accustomed to having conversations with machines. In travel, we have been experimenting with conversational interfaces for years. I remember trying quite a few amazing tools back as far as five years ago, but we could never convince the average customer to type more than a word, such as a destination name, into the text boxes. The popularity of platforms like TikTok, YouTube Shorts and Tinder shows how the average customer is drawn to being fed information fast, at the click of a button, rather than having a lengthy conversation. Maybe there is a way to combine these experiences, but it is a big open question.
On the other hand, the buzz around generative AI may be just what we need to finally encourage consumers to use a new conversational, yet very intuitive interface. The potential disruption here is high. But the same was said about the world moving towards Alexa and voice interfaces, and we haven’t seen that come into effect yet. Not to mention the metaverse or Crypto.
New search engine or consumer marketing channels
I am on the skeptical side on this one. While there are definitely use cases where conversational is better than the typical search experience, I think those are not the majority. Google is forced to explore evolving its UX and introduce more conversational elements into search, and I think they will do so successfully eventually (Bard will get there). To some extent this already exists with the short answers Google gives you above the search results.
It may be just beyond the medium future, but I think there’s much more disruption potential in the next generation of personal assistants, tools like AutoGPT and their evolutions. If you read the materials of the famous psychologist Daniel Kahneman, he described the human brain as two systems:
- System 1 is intuitive, converges quickly, is fast to judge in absolute terms, but tends to make mistakes – what we sometimes call biases. Sounds familiar? This is how you do one-digit multiplications and assess the danger on a road looking at it. Computers already could be a much better system 1 than the human brain when it comes to numbers. Now for the first time, they’re also much better at language and written skills.
- System 2 requires energy and grit. It allows us to solve harder problems by breaking it into small components, by coming up with strategies, through trial and error, by doubting ourselves, etc. This is how most of us do 3-digit multiplications, write an essay, or engage in a thoughtful debate. AI can do some of what our human brain does within this system, some of it as intuitively as system 1, but a lot is still missing.
Tools like AutoGPT realize that a human is simply a combination of these two systems and some working memory. So new algorithms come up that convert and execute complex challenges into smaller tasks, that strategize and even self-correct the mistakes of the “simpler though definitely not simple” first attempt. While those AI skills are still in the infantry stages, these in combination with personalization may change the world. After all, do you really need a conversation with your personal assistant (or travel agent) if they know you so well that they’ll figure it out for you, and probably better than you would yourself? Maybe all you need to do is give them your destination, occupancy, and click search & book. They’ll call you if they need your help.
Whatever the future holds, I am in the camp of optimists, and am convinced of disruption. It’s already happening, and now we’ll see if it will find its way from the backend to the consumer interfaces, and to what extent. That for me is the big question. I can’t wait to find out and am even more excited to potentially be a force in shaping the outcome.
This article was originally published by Omri Morgenshtern, Chief Executive Officer at Agoda, on LinkedIn
