Digital Innovation
“I
n the future, there will
be two types of job,”
forecasts Mark Brayan,
CEO of Sydney-based
software company Appen. “The ones
where you tell computers what to do
and the ones where computers tell you
what to do.
“Be accepting,” he advises.
“Automation is going to enhance
work, disrupting some jobs and
creating others, but it will make a lot
more people more effective. It’s the
professions that don’t embrace it that
will be left behind.”
There is no danger of Luddism
at Appen; tech lies at its heart.
Headquartered in Australia, but
with offices in the US, UK and the
Philippines, it is a truly global and
thoroughly 21st century organisation:
artificial intelligence (AI) is its bread
and butter, and the gig economy a core
element of its recruiting model.
“Our environment is unique from
a work perspective,” explains Brayan.
“The gig economy is evolving and we
offer a case study from that perspective.
But we’re also in this world of AI, which
is disrupting work too.”
speech data and improving the way
humans interact with computers in
natural languages. But over the past
five years, we’ve started gathering larger
datasets consistent with the onset of AI
and machine learning.”
In lay terms, Appen’s raison d’être
is to train robots – and its clients
comprise some of the world’s largest
global technology firms (including
Microsoft) plus the auto sector and
several governments.
“Like people, robots learn from
information and experience, but that
information and experience is data,”
says Brayan. “So if you want to train a
bot or programme to mimic a human
function such as speech, sight or
making a decision, you feed it related
data and it enables the software to build
a model that does that.
“There’s a surprising amount of this
in everyday products such as in search
engines and social media. They use a
branch of computing called machine
learning. This requires data, and that’s
what we provide.”
If this sounds cold and automated,
devoid of all human input, it couldn’t
be further from the truth.
“The interesting thing, from a
recruiting perspective, is that we rely on
humans to do this,” says Brayan. “If you
want to get a product to mimic a human
function, you need humans to collect,
annotate or enhance that data in a way
that gives the computer the human’s
version of that. We’re in AI but we’re
creating jobs rather than killing them.”
Applying the
human touch
Appen’s permanent workforce
comprises 330-plus members (largely
project managers, linguists, customer
service people and engineers), plus
a crowd of 400,000 people on whom
it relies for data collection and
enhancement. The company has
worked in 130 countries, in more than
180 languages.
“We may pay 15,000-20,000 people
in any given month to do work for
us, mostly from home,” says Brayan.
“These contractors work on the data
– for example, assessing the relevance
of a web search. This is all done online
and delivered to our clients. If we need
people who speak Turkish, Brazilian or
The evolution of AI
Initially focused on providing language
and linguistic consulting to tech
companies, Appen is now a global leader
in the development of high-quality,
human-annotated data for machine
learning. Brayan explains: “We’ve been
in this area for over 20 years, collecting
“In the future, there will be two types of job: the
ones where you tell computers what to do, and
the ones where computers tell you what to do”
Issue 2 - 2017
51