In recent months, numerous products touting the use of
artificial intelligence (AI) in the form of chatbots or voice interfaces have
come online in the travel space, signaling that the technology behind them has
advanced to the point where interacting with machines has become useful.
Even so, while travel companies continue to invest in new AI
technologies to help streamline operations and interact with clients, those
technologies have not advanced to the point where they could replace humans,
said Norm Rose, senior technology and corporate market analyst with Phocuswright.
This July, within the space of days, two new travel chatbots
were announced. Mezi, a virtual travel assistant powered by AI, launched Mezi
for Business and announced clients that included Adelman Travel and Casto Travel. Sabre said it was testing an AI-powered chatbot with Travel Solutions
International USA and Casto.
That same week, Kayak launched the ability to book hotel
rooms using the Amazon Echo device and its virtual personal assistant, Alexa;
previously, the feature merely offered search functions. A month later, booking
ability was extended to the Echo Show, a device with a screen that enables
visual elements. Similar voice features are available on other devices, such as
Google Home.
In recent weeks, the travel club Mr & Mrs Smith
introduced a voice-activated search of its hotel inventory via Alexa, NYC &
Company implemented an AI chatbot as a resource for visitors, and Cruise
Planners announced both consumer- and agent-facing Alexa skills, among other
recently introduced products.
"The technology has come along," Rose said, which
is why so many new products are being introduced.
Facets of AI have been in existence in some travel brands
for years, according to Matthias Keller, Kayak's chief scientist, who oversees
the company's AI efforts. For example, machine learning has long influenced the
sort order of hotels on Kayak.
"But now, with the chatbots and the voice agents,
machine learning/AI is something that you can touch and feel, so to speak,"
Keller said. "It's coming to the surface, and it's recognized by everyone
as a major driver of this whole innovation, this whole new space."
The quality of chatbots and voice recognition technology has
been steadily improving as well, he said. Five years ago, speaking to AI
technology was like speaking to a robot, but new technologies are overcoming
that hurdle, he said.
While recent advances have been numerous, chatbots have been
around for a long time. Rose covered their history in a 2016 Phocuswright
report titled "Travel chatbots are hot, but will they replace or augment
human conversation?"
The concept of chatbots began in 1966, when an MIT scientist
created a program called Eliza that mimicked human conversation. The first wave
of chatbots appeared in the 1990s, but answers that had not been anticipated
tended to hit a dead end, and the chatbots failed to be conversational.
Today, Rose said, "we have reached a point where being
able to use semantic understanding, natural language processing, to have a
conversational interface via chat has matured [to the point] that it does
handle some things. But it is a mistake for the market to think it is a human
replacement; it has limited capabilities."
Those capabilities, however, lend themselves well to an
environment like corporate travel, as evidenced by Mezi's and Sabre's efforts
in the area.
Sabre said its chatbot is designed to handle frequent, basic
service and support requests like changing a flight. More complex requests are
routed to a human agent.
Mezi for Business operates largely the same way. Moreover,
using Mezi's platform, agents are able to train the AI to deal with requests it
has failed to fulfil so that it can answer similar requests in the future. CEO
Swapnil Shinde said the AI answers about 60% of requests now, and the company
hopes to increase that to 80% in the next three to six months.
Travel management companies are attracted to a product like
Mezi because it offers a "high-touch, personal experience" to
travelers, Shinde said, and it helps agents become more productive. Marco,
Casto Travel's white-labeled app powered by Mezi, has already made its agents
five times more productive, he said.
"Imagine a travel agent who was doing 1,000 bookings a
month," Shinde said. "Now, the same travel agent can do 5,000
bookings a month."
Productivity is also increased by lessening training time.
Shinde said an agent can be fully trained on the Mezi dashboard in three to
five days.
Marc Casto, president and CEO of Casto Travel, was quick to
jump on board new technologies such as Sabre's and Mezi's chatbots. Portions of
Casto's corporate client base prefer quick turnaround times that bots provide
in many cases, he said, and the conversational interface is more appealing than
forms on a website, for example.
"It's quick, it's reliable, and it's much more
intuitive than most communications when using automated systems," he said.
Casto agreed that AI technology is improving, but its cost
is also coming down, an important consideration for agencies like his.
"It's possible for all of us to start approaching
these," he said. "Even if we're looking at 12 months back, the cost
of developing and maintaining your own chatbots was pretty significant."
Rose, the Phocuswright analyst, cautioned that chatbots face
some hurdles. For one thing, not everyone wants to use a chat interface. What's
more, unlike frequent corporate travelers, leisure travelers typically lack a
detailed profile for a chatbot to pull data from when making product
recommendations.
AI-powered voice technology is likely the "more
dramatic disruption to the way things are done today," Rose said, because
it enables users to simply speak, instead of type what they want. It does face
challenges, though.
For example, if a user is looking for the lowest-cost flight
from San Francisco and Boston, how does the technology define lowest? Would it
give the user a result that includes a ridiculously long layover?
Also, planning travel requires information. Chatbots tend to
ask a number of questions, Rose said, but at what point would a voice interface
user become frustrated and switch to a different planning method because voice
is no longer providing efficiency?
"I think we're at the early stages of all this stuff,"
Rose said. "And while it shouldn't be ignored, it also shouldn't be viewed
as the future that everyone is going to be participating in. It's going to be
for sectors of the population," likely younger generations.
Keller acknowledged that Kayak is still in the early stages
of offering voice-enabled hotel bookings, but consumers have been using the
feature.
"People are very excited that this is happening,"
he said. He admitted that some might not be ready to fully trust a voice
interface, but predicted that, "within the next year, this [technology] is
going to be a real e-commerce channel and a real channel for users to regularly
interact with our brand."
Keller is a believer in voice in travel because talking is
simply easier than typing for most people. He said Kayak will continue
investing in the technology, hopefully enabling users to book more things, like
flights, via voice.
"There are a lot of things missing right now," he
said, but time and money is being invested to solve those problems.
Rose and Casto both asserted that while technologies like
voice and chatbots are advancing, they are still in their infant stage, with
growth to come.
"I personally believe that we're in the equivalent of
the AOL, 'You've got mail' stage of Internet production as it relates to
robotics," Casto said. "We are [so] at the early stage of what we're
going to see -- how this impacts travel -- that it's very hard to anticipate
what the future holds. But the lesson learned from that environment is that I
truly believe it will be all-encompassing, it will be unavoidable, and it would
be foolish to not be making investments in this area."