Monday5 August 2019 ★ FINANCIAL TIMES 7COMPANIES & MARKETS
RICHARD WATERS— SAN FRANCISCOIn the race to build a machine with
human-level intelligence, it seems, size
really matters.
“We think the most benefits will go to
whoever has the biggest computer,” said
Greg Brockman, chairman and chief
technology officer of OpenAI.
The San Francisco-based AI research
group, set up four years ago by tech
industry luminaries including Elon
Musk, Peter Thiel and Reid Hoffman,
has just thrown down a challenge to the
rest of the AI world.
Late last month, it raised $1bn from
Microsoft to speed its pursuit of the
Holy Grail of AI: a computer capable of
so-called artificial general intelligence
(AGI), a level of cognition that would
match its makers, and which is seen as
the final step before the advent of com-
puters with superhuman intelligence.
According to Mr Brockman, that
money — a huge amount for a research
organisation — will be spent “within five
years, and possibly much faster”, with
the aim of building a system that can
run “a human brain-sized [AI] model”.
Whether a computer that matches
the neural architecture in the human
brain would deliver a comparable level
of intelligence is another matter.
Mr Brockman is wary about predict-
ing precisely when AGI will arrive, and
said that it would also require advances
in the algorithms to make use of the
massive increase in computing power.
But, speaking of the vast computing
power that OpenAI and Microsoft hope
to put at the service of its AI ambitions
within five years, he added: “At that
point, I think there’s a chance that will
be enough.”
OpenAI’s huge bet points to a parting
of the ways in the artificial intelligence
world after a period of rapid advance.
Deep learning systems, which use artifi-
cial neural networks modelled on one
idea of how the human brain works,
have provided most of the break-
throughs that have put AI back at the
centre of the tech world. OpenAI argues
that, with enough computing power,
there is a good chance these networks
will evolve further, right up to the level
of human intelligence.
But many AI researchers believe deep
learning on its own will never become
much more than a form of sophisticated
pattern-recognition — perfect for facial
recognition or language translation, but
far short of true intelligence.
Some of the most ambitious research
groups — including DeepMind, the Brit-
ish AI research company owned by
Alphabet — believe that teaching com-
puters new types of reasoning and sym-
bolic logic will be needed to comple-
ment the neural networks, rather than
just building bigger computers.
“If we allocated $100m for compute,
what could we do? We’re thinking about
it, and you can imagine other people are
thinking about it as well,” said Oren Etz-
ioni, the head of Allen Institute for Arti-
ficial Intelligence, one of the best-
funded American AI research groups.
But he added: “To reach the next level
of AI, we need some breakthroughs. I’m
not sure it’s simply throwing more
money at the problem.”
Others are more forthright. Asked
whether bigger computers alone will
deliver human-level AI, Stuart Russell, a
computer science professor at the Uni-
versity of California, Berkeley, points to
the verdict in his forthcoming book on
the subject: “Focusing on raw comput-
ing power misses the point
entirely... We don’t know how to
make a machine really intelligent —
even if it were the size of the universe.”
Even the possibility that OpenAI may
be on the right track, though, has been
enough to attract a huge cash injection
from the world’s most valuable com-
pany, setting up a race to build far more
advanced hardware systems for AI. Mr
Brockman calls it “a public benefitApollo program to build general intelli-
gence”. That reflects the mission set by
OpenAI’s founders, to build an AI whose
benefits are not limited to one corpora-
tion or individual government.
It may also create unmatched wealth.
Pointing to the market value of today’s
top tech companies, he said: “That’s the
value we produce with computers that
aren’t very smart. Now imagine we suc-
ceed in building the kind of technology
we’re talking about, an artificial general
intelligence — that company is going to
be by a huge margin unprecedented in
history, the number one.”
OpenAI’s bet is that, as computer
hardware gets more powerful, the learn-
ing algorithms used in deep learning
systems will evolve, developing capabil-
ities that today’s coders could never
hope to program into them directly.
It is a controversial position. Critics
like Mr Russell argue that simply throw-
ing more computing power at imperfect
algorithms means “you just get the
wrong answer more quickly”. Mr Brock-
man’s response: “You can get qualita-
tively different outcomes with
increased computation.”
He claims some of the tests carried
out by OpenAI in its four-year history
hint at the kind of advances that could
come from massive increases in hard-
ware. Two years ago, for instance, the
researchers reported the results of a sys-
tem that read customer reviews on
Amazon and then used statistical tech-
niques to predict the next letter.
The system went further, according to
OpenAI, learning for itself the differ-
ence between positive and negative sen-
timent in the review — a level of under-
standing beyond anything that might
have been expected. One of OpenAI’s
recent experiments — an AI system thatbeat a top human team at the video
gameDota 2— also showed that today’s
most advanced AI systems can perform
well at games that are far closer to the
real world than board games like chess.
That echoed work by DeepMind on
playing the gameStarcraft. According to
Mr Brockman, the OpenAI system
taught itself to operate at a higher level
of abstraction, setting an overall goal,
then “zooming in” on particular tasks as
needed — the kind of planning that is
seen as a key part of human intelligence.
Even many of the scepticsseem wary
of writing off its claims completely. “It’s
fair to say that deep learning has been a
paradigm shift,” said Mr Etzioni. “Can
they achieve something like that again?”
Bringing in Microsoft is a change in
direction for the research group. Most of
the $1bn investment will return to the
software company inpayments to use
its Azure cloud computing platform,
with Microsoft working on new super-
computing capabilities to aid the effort.
Mr Brockman denies this is a deviation
from OpenAI’s goal of staying above the
corporate fray. Microsoft, he said, would
be limited to the role of “investor and a
strategic partner in building large-scale
supercomputers together”.
The software company’s investment
will give it a large minority stake in
OpenAI’s for-profit arm, as well as one
seat on its board. Like all of the organi-
sation’s equity investors, its potential
returns have been capped at a fixed
level, which has not been disclosed.
If OpenAI’s work ever produces the
huge wealth Mr Brockman predicts,
most of it will flow to the group’s non-
profit arm, reflecting its promise to use
the fruits of advanced computer intelli-
gence for the benefit of all humanity.
See LexThe tech industry is accustomed to
riding the curve of Moore’s Law,
which describes the way that
computing power roughly doubles
every two years. But OpenAI is
counting on a more powerful
exponential force to take the
capacity of its AI systems to a level
that seems almost unimaginable
today. The group calculates that
since the tech industry woke up to
the potential of machine learning
seven years ago, the amount of
processing capacity being applied
to training the biggest AI models
has been increasing at five times
the pace of Moore’s Law.
That makes today’s most
advanced systems 300,000 times
more powerful than those used in- The advance reflects the
amount of money being poured
into advanced AI, as well as the
introduction of parallel computing
techniques that make it possible to
crunch far more data.
Mr Brockman said OpenAI is
counting on this exponential trend
being carried forward another five
years — something that would
produce results that, he admits,
sound “quite crazy”. As a
comparison, he said that the last
seven years of advances would be
like extending the battery life of a
smartphone from one day to 800
years: another five years on the
same exponential curve would take
that to 100m years. Today’s most
advanced neural networks are
roughly on a par with the honey
bee. But with another five years of
advances, OpenAI believes it has a
shot at matching the human brain.
Richard Waters
Brain game
Pace of advances likely
to end in ‘crazy’ resultsThe billion-dollar bet to reach human-level AI
With backing from Microsoft, San Francisco-based research group aims to create intelligent machines within 5 years
ALICE HANCOCK— LONDONJuul, the US start-up, launched the first
in a series of bluetooth-connected
e-cigarettes that can monitor users’
vaping and track their device as it seeks
to stamp out fears around take-up of
the activity among teenagers.The Juul C1, as the device is known,
started selling in the UK this week after
a successful pilot in Canada.
Dan Thomson, managing director of
Juul’s UK business, said that in order to
use the device, which is linked to an app
on users’ smartphones, customers had
to go through stringent age verification
checks that included facial recognition
and a two-step background check with
third-party databases.Juul has come under fire in the US for
its high take-up among teenagers, which
has prompted a crackdown on vaping
by the US Food and Drug Administra-
tion. Last year the organisation said the
problem had reached “epidemic” levels.
Retailers are required by a Juul policy
to ID-check anyone who looks younger
than 25. It is illegal to own a vape in the
UK under the age of 18. In June, San
Francisco, where Juul is based, became
the first US city to ban e-cigarettes in a
bid to curb youth uptake.
Juul’s C1 device will allow users to
monitor how many puffs they take a
day, as well as locate their vape if they
lose it. It also has an auto-lock option
that means the e-cigarette can be locked
when out of the phone’s range in orderto prevent others using it. Roxy Wacyk,
director of product management at
Juul, said the company had a “regular
cadence of app updates” planned and
that they were looking at features that
allowed users to limit their vaping, and
geofencing against usage in public areas
such as schools.
However, privacy experts have raised
concerns about the gathering of data
required to develop the device.
“Data about the machine that goes
back to the manufacturer can tell you a
lot about the user and who that user is,”
said Martin Garner, chief operating
officer of the technology consultancy
CCS Insight.
Juul said it had “no plans at this time”
to sell any data to other organisations.Tobacco
Juul launches e-cigarette to track users’ vaping
OpenAI hopes that algorithms will evolve beyond the level humans could program into them as computer hardware becomes more powerful— Alamy
РЕ
ЛИЗП
Пand a two-step background check withand a two-step background check withОchecks that included facial recognition
Оchecks that included facial recognition
and a two-step background check withand a two-step background check withОchecks that included facial recognitionchecks that included facial recognitionДД
and a two-step background check withД
and a two-step background check withchecks that included facial recognitionchecks that included facial recognitionГГОto go through stringent age verification
Оto go through stringent age verification
checks that included facial recognitionchecks that included facial recognitionОto go through stringent age verificationto go through stringent age verificationТТОon users’ smartphones, customers had
Оon users’ smartphones, customers had
to go through stringent age verificationto go through stringent age verificationОВon users’ smartphones, customers had
Вon users’ smartphones, customers had
to go through stringent age verificationto go through stringent age verificationВon users’ smartphones, customers hadon users’ smartphones, customers hadИЛuse the device, which is linked to an app
Лuse the device, which is linked to an app
on users’ smartphones, customers hadon users’ smartphones, customers hadЛuse the device, which is linked to an appuse the device, which is linked to an appАА
on users’ smartphones, customers hadА
on users’ smartphones, customers haduse the device, which is linked to an appuse the device, which is linked to an appГГРJuul’s UK business, said that in order to
РJuul’s UK business, said that in order to
use the device, which is linked to an appuse the device, which is linked to an appРJuul’s UK business, said that in order toJuul’s UK business, said that in order toJuul’s UK business, said that in order toJuul’s UK business, said that in order toУУПП
ПDan Thomson, managing director of
ПDan Thomson, managing director of
Juul’s UK business, said that in order toJuul’s UK business, said that in order toПDan Thomson, managing director ofDan Thomson, managing director ofА"What'sto go through stringent age verification"What'sto go through stringent age verification
checks that included facial recognition
"What'schecks that included facial recognition
and a two-step background check withand a two-step background check with "What'sNews"on users’ smartphones, customers had
News"on users’ smartphones, customers had
to go through stringent age verificationto go through stringent age verificationNews"VK.COM/WSNWSon users’ smartphones, customers hadVK.COM/WSNWSon users’ smartphones, customers had
to go through stringent age verificationVK.COM/WSNWSto go through stringent age verification
checks that included facial recognitionVK.COM/WSNWSchecks that included facial recognition
and a two-step background check with
VK.COM/WSNWSand a two-step background check with
third-party databases.third-party databases.VK.COM/WSNWS