At a time when big data and artificial intelligence (AI) are
becoming increasingly important elements of evolving travel technologies, they
are being called out as culprits in the aftermath of the presidential election
after every major forecast called for a Hillary Clinton victory.
But industry experts, while citing several lessons to be
learned from the flawed polling results, insisted last week that big data
itself was not to blame. Nor, they said, was AI that in a growing number of
cases is used to process big data.
The likely cause of so many forecasts erroneously in Clinton's
favor, they said, was the flawed ways in which much of the polling data was
gathered. For that reason, they added, it should not give travel companies
pause about using big-data-based marketing solutions.
"Big data itself is not to blame," said analyst
Henry Harteveldt, founder of Atmosphere Research Group. "When it comes to
political polling vs. commercial market research and the use of big data and so
on, there are a couple of things to bear in mind."
First, he said, political polls are largely dependent on who
answers their phones, what numbers are called -- pollsters without access to
mobile phones likely missed a large subset of the country that no longer has a
landline -- and the willingness of a respondent to report opinions candidly.
Harteveldt also said there was some "social stigma"
at play, with some Donald Trump voters probably being unwilling to acknowledge
their support of him.
Norm Rose, senior technology and corporate market analyst at
Phocuswright, agreed that big data was not to blame in election-outcome
predictions.
"In this case, the presidential election seems like it
was a failure in polling," he said. "It doesn't tarnish, in my mind,
the idea of using big data to understand your [travel] customers better."
Even so, Harteveldt said, "For anyone in travel who is
[using] or thinking of using AI, there are lessons to be learned in what to do
and what not to do from the political campaigns. But they are not a blueprint
to success, nor are they case studies to use so that you avoid making some of
the mistakes."
In Harteveldt's mind, the key lesson to be learned is that "data
can guide you, but it cannot make the decisions for you. You have to interpret
the data correctly."
He said it is important to keep in mind that even with data
there is some subjectivity. The right data needs to be collected in the first
place.
"There's a lot of internal information that can be
collected, but there's important external information that also needs to be
gathered and aggregated and analyzed," he said.
Travel-tech companies that employ big data and/or AI (facets
of artificial intelligence such as machine learning frequently parse big data)
indeed have seen some lessons to be learned from the election.
"There's this idea with AI and big data that you can
take a hands-off approach and let it work its magic, when in fact, humans are
still needed across industries to read between the lines," said Swapnil
Shinde, CEO and co-founder of Mezi, a personal assistant that uses AI to book
travel.
"Even relationships with chatbots that center around
planning travel are built on trust in not just the technology but also in the
humans who built the technology," he said. "This is a good reminder
to those working in industries that rely on AI and machine learning to not
forget the human element of big data and AI."
Devon Tivona, the CEO of Pana -- an app that augments travel
agents' abilities with an assist from AI -- said he believes big data has a
place in the travel industry. In fact, he called himself "bullish" on
the impact it could have. "However, the lesson to be learned here is that
we have to be careful with our data," Tivona said. "We need to ensure
that we're making proper inferences from the data that we're seeing."
Gillian Morris is the founder of Hitlist, an app that uses
big data and personalization to alert users when cheap flights are available
for trips they want to take. Morris said that when considering large amounts of
data, it's possible to put too much stock in what the data said and not enough
into personal context and social recommendations.
For example, technology could identify a hotel that,
statistically, is 95% likely to be the right fit for a consumer, but that
consumer might have had a friend stay there who'd had a bad experience, making
that particular hotel an unlikely choice for a reason not included in the data.
"If the travel industry is to draw conclusions from the
election upset," she said, "it may simply be that we're not that
smart yet. The best engineering teams and product machines can and will be
disrupted by a flawed product that resonates with everyday consumers."
Travel technologies that employ both big data and facets of
AI are still in their early days, but Rose said the election shouldn't be
off-putting to those who are developing them. "I think we're at an early
stage here, and I think it's going to develop," he said, "but I don't
look at the election as a repudiation of big data."