Apple’s new AI acquisition will help it make sense of data

screen-shot-2017-05-13-at-13-56-34 Dark data isn’t as sinister as it sounds.
Photo: Lattice Data

Apple has acquired an AI company as part of its continued push
to embrace artificial intelligence.

The company in
question is the Menlo Park-based Lattice Data, which specializes in
taking unstructured, “dark” data and transforming it into more
useful, structured information. Apple acquired around 20
engineers as part of the deal.

According to TechCrunch
who reported the buyout, the price Apple paid for Lattice Data
is in the realm of $200 million. The deal was concluded a
couple of weeks ago, and in a statement from Apple, it gave its
usual stock explanation for the acquisition: “Apple buys
smaller technology companies from time to time and we generally
do not discuss our purpose or plans.”

Exactly what Apple has planned for the
acquisition or its new engineers isn’t clear. However, the job
of turning unstructured data into something that can be used by
machine learning systems is something AI researchers continue
to work on.

explanation that TechCrunch puts forward is that Apple
will be using Lattice Data’s engineers to work on Siri, which
would make sense given that it is reportedly planning to

launch a standalone Siri speaker
in the near future —
thereby making it more central to Apple’s product offerings.
Apple previously switched over to
using deep learning technology
in July 2014.

Going big on AI

lagging behind in AI research for a long time, Apple has been
ramping up its investment in the field
over the past several years
. Late last year, Apple
researchers published the
company’s first ever research paper
, describing a method
for training AI algorithms to recognize images.

This move
is one of many Apple has taken to make it more appealing to AI
researchers. A

suggests that Apple’s user privacy policy has gotten
in the way of it recruiting some AI students — since they
want access to the kind of data Apple doesn’t collect about its

%d bloggers like this: