FIVE OF THE SMARTEST ROBOTS IN HISTORY
One of the first examples
of a chatterbot. When
running a script dubbed
DOCTOR, ELIZA
could ask and answer
questions like a
psychotherapist. It didn’t
understand a great deal,
but, with some clever
programming, was
still able to convince
many users of its
intelligence.1966
ELIZAComputer artist and
researcher Karl Sims
created a group of
virtual creatures that
inhabited their own
virtual universe. Using
genetic algorithms,
they evolved until they
could swim, crawl,
jump and compete
against one another.
Unfortunately, they
were too concerned
with their own virtual
lives to talk to us.1994
VIRTUAL
CREATURES
Deep Blue was
catapulted into public
consciousness when
it won a chess game
against grandmaster
Garry Kasparov –
the first supercomputer
to achieve such a feat.
However, it was given
a lot of help from human
programmers and used
pretty basic AI methods
to think of its moves,
so maybe it was not
so bright after all.1997
DEEP BLUEIBM Watson was the first
AI to beat human players
at the US TV quiz show
Jeopardy!. This AI was
clever enough to
process text and then
found likely answers to
the questions asked
using its internal body
of knowledge, which
comprised around 200
million pages of content.
Sadly, it struggled to
answer some basic
questions.2011
IBM WATSONThis earlier work
from the team behind
AlphaGo, the AI that
defeated a master of
the complex Japanese
strategy game Go,
learned how to play 49
classic Atari games just
by looking at the screen- it didn’t get any help
from programmers.
While it was brilliant
at a lot of the games,
it couldn’t get the hang
of Pac-Man.
2015
DEEP-Q NETWORK“I think in three to five
years we will forget
what it was like when
our devices didn’t
understand emotion”
emotions from our faces. Their AI is trained on a vast database
of more than half a million faces analysed from people in
75 countries, with 50 million new emotion data points –
a face expressing emotion such as happiness, sadness or surprise –
being added every day.
“We are giving machines the ability to sense and respond to
human emotion, something that is deeply human but that
today’s technology has not been capable of doing,” says Kaliouby.
“We like to say we are bringing AI to life!”
Tomorrow’s robots will not be mere machines, cold
and heartless. They will be emotionally aware – and it will
happen soon, researchers say.
“I think, in three to five years, we will forget what it was like
when our devices didn’t understand emotion,” says Kaliouby.
“It’s similar to how we all assume that our phones today are
location-aware. Someday soon, it will be the same for emotions.”
Tresset wonders what the robots that learn will be able to do
in the future. “Robots can already learn, but, as long as they are
not able to take the decision to produce art, they cannot be seen
as artists. Intentionality is very important in art,” he says.
“If a robot on an assembly line starts to hit a car to produce
a sculpture, then will it be an artist? If a military drone starts to
dance in the sky, will it be an artist?”
“When we design intelligent robots,” says Aloimonos, “it is as
if we are trying to understand ourselves – it is what the ancient
Greeks referred to as ‘gnothi seauton’ [‘know thyself’].
This quest will never end.”Dr Peter Bentley is a computer scientist and author based at University
College London.