<s>
Artificial	B-Application
intelligence	I-Application
applications	I-Application
have	O
been	O
used	O
in	O
a	O
wide	O
range	O
of	O
fields	O
including	O
medical	O
diagnosis	O
,	O
stock	O
trading	O
,	O
robot	O
control	O
,	O
law	O
,	O
scientific	O
discovery	O
,	O
video	B-Application
games	I-Application
,	O
and	O
toys	O
.	O
</s>
<s>
However	O
,	O
many	O
AI	B-Application
applications	I-Application
are	O
not	O
perceived	O
as	O
AI	B-Application
:	O
"	O
A	O
lot	O
of	O
cutting	O
edge	O
AI	B-Application
has	O
filtered	O
into	O
general	O
applications	O
,	O
often	O
without	O
being	O
called	O
AI	B-Application
because	O
once	O
something	O
becomes	O
useful	O
enough	O
and	O
common	O
enough	O
it	O
's	O
not	O
labeled	O
AI	B-Application
anymore.	O
"	O
</s>
<s>
"	O
Many	O
thousands	O
of	O
AI	B-Application
applications	I-Application
are	O
deeply	O
embedded	O
in	O
the	O
infrastructure	O
of	O
every	O
industry.	O
"	O
</s>
<s>
In	O
the	O
late	O
1990s	O
and	O
early	O
21st	O
century	O
,	O
AI	B-Application
technology	O
became	O
widely	O
used	O
as	O
elements	O
of	O
larger	O
systems	O
,	O
but	O
the	O
field	O
was	O
rarely	O
credited	O
for	O
these	O
successes	O
at	O
the	O
time	O
.	O
</s>
<s>
Kaplan	O
and	O
Haenlein	O
structure	O
artificial	B-Application
intelligence	I-Application
along	O
three	O
evolutionary	O
stages	O
:	O
1	O
)	O
artificial	O
narrow	O
intelligence	O
–	O
applying	O
AI	B-Application
only	O
to	O
specific	O
tasks	O
;	O
2	O
)	O
artificial	O
general	O
intelligence	O
–	O
applying	O
AI	B-Application
to	O
several	O
areas	O
and	O
able	O
to	O
autonomously	O
solve	O
problems	O
they	O
were	O
never	O
even	O
designed	O
for	O
;	O
and	O
3	O
)	O
artificial	O
super	O
intelligence	O
–	O
applying	O
AI	B-Application
to	O
any	O
area	O
capable	O
of	O
scientific	O
creativity	O
,	O
social	O
skills	O
,	O
and	O
general	O
wisdom	O
.	O
</s>
<s>
To	O
allow	O
comparison	O
with	O
human	O
performance	O
,	O
artificial	B-Application
intelligence	I-Application
can	O
be	O
evaluated	O
on	O
constrained	O
and	O
well-defined	O
problems	O
.	O
</s>
<s>
This	O
gives	O
better	O
insight	O
into	O
the	O
comparative	O
success	O
of	O
artificial	B-Application
intelligence	I-Application
in	O
different	O
areas	O
.	O
</s>
<s>
AI	B-Application
,	O
like	O
electricity	O
or	O
the	O
steam	O
engine	O
,	O
is	O
a	O
general	O
purpose	O
technology	O
.	O
</s>
<s>
There	O
is	O
no	O
consensus	O
on	O
how	O
to	O
characterize	O
which	O
tasks	O
AI	B-Application
tends	O
to	O
excel	O
at	O
.	O
</s>
<s>
While	O
projects	O
such	O
as	O
AlphaZero	B-Application
have	O
succeeded	O
in	O
generating	O
their	O
own	O
knowledge	O
from	O
scratch	O
,	O
many	O
other	O
machine	O
learning	O
projects	O
require	O
large	O
training	O
datasets	O
.	O
</s>
<s>
Researcher	O
Andrew	O
Ng	O
has	O
suggested	O
,	O
as	O
a	O
"	O
highly	O
imperfect	O
rule	O
of	O
thumb	O
"	O
,	O
that	O
"	O
almost	O
anything	O
a	O
typical	O
human	O
can	O
do	O
with	O
less	O
than	O
one	O
second	O
of	O
mental	O
thought	O
,	O
we	O
can	O
probably	O
now	O
or	O
in	O
the	O
near	O
future	O
automate	O
using	O
AI.	O
"	O
</s>
<s>
AlphaGo	B-Application
brought	O
the	O
era	O
of	O
classical	O
board-game	O
benchmarks	O
to	O
a	O
close	O
when	O
Artificial	B-Application
Intelligence	I-Application
proved	O
their	O
competitive	O
edge	O
over	O
humans	O
in	O
2016	O
.	O
</s>
<s>
Deep	B-Application
Mind	I-Application
’s	I-Application
AlphaGo	B-Application
AI	B-Application
software	I-Application
program	O
defeated	O
the	O
world	O
’s	O
best	O
professional	O
Go	O
Player	O
Lee	O
Sedol	O
.	O
</s>
<s>
Games	O
of	O
imperfect	O
knowledge	O
provide	O
new	O
challenges	O
to	O
AI	B-Application
in	O
the	O
area	O
of	O
game	O
theory	O
;	O
the	O
most	O
prominent	O
milestone	O
in	O
this	O
area	O
was	O
brought	O
to	O
a	O
close	O
by	O
Libratus	O
 '	O
poker	O
victory	O
in	O
2017	O
.	O
</s>
<s>
E-sports	O
continue	O
to	O
provide	O
additional	O
benchmarks	O
;	O
Facebook	B-Application
AI	B-Application
,	O
Deepmind	B-Application
,	O
and	O
others	O
have	O
engaged	O
with	O
the	O
popular	O
StarCraft	B-Application
franchise	O
of	O
videogames	O
.	O
</s>
<s>
Broad	O
classes	O
of	O
outcome	O
for	O
an	O
AI	B-Application
test	O
may	O
be	O
given	O
as	O
:	O
</s>
<s>
Optical	B-Application
character	I-Application
recognition	I-Application
for	O
ISO	O
1073-1:1976	O
and	O
similar	O
special	O
characters	O
.	O
</s>
<s>
Various	O
robotics	O
tasks	O
that	O
may	O
require	O
advances	O
in	O
robot	O
hardware	O
as	O
well	O
as	O
AI	B-Application
,	O
including	O
:	O
</s>
<s>
The	O
Feigenbaum	O
test	O
,	O
proposed	O
by	O
the	O
inventor	O
of	O
expert	B-Application
systems	I-Application
,	O
tests	O
a	O
machine	O
's	O
knowledge	O
and	O
expertise	O
about	O
a	O
specific	O
subject	O
.	O
</s>
<s>
A	O
paper	O
by	O
Jim	O
Gray	O
of	O
Microsoft	O
in	O
2003	O
suggested	O
extending	O
the	O
Turing	O
test	O
to	O
speech	B-Application
understanding	I-Application
,	O
speaking	B-Application
and	O
recognizing	O
objects	O
and	O
behavior	O
.	O
</s>
<s>
At	O
an	O
extreme	O
,	O
the	O
test	O
suite	O
can	O
contain	O
every	O
possible	O
problem	O
,	O
weighted	O
by	O
Kolmogorov	O
complexity	O
;	O
however	O
,	O
these	O
problem	O
sets	O
tend	O
to	O
be	O
dominated	O
by	O
impoverished	O
pattern-matching	O
exercises	O
where	O
a	O
tuned	O
AI	B-Application
can	O
easily	O
exceed	O
human	O
performance	O
levels	O
.	O
</s>
<s>
Many	O
competitions	O
and	O
prizes	O
,	O
such	O
as	O
the	O
Imagenet	B-General_Concept
Challenge	I-General_Concept
,	O
promote	O
research	O
in	O
artificial	B-Application
intelligence	I-Application
.	I-Application
</s>
<s>
The	O
most	O
common	O
areas	O
of	O
competition	O
include	O
general	O
machine	B-Application
intelligence	I-Application
,	O
conversational	O
behavior	O
,	O
data-mining	O
,	O
robotic	O
cars	O
,	O
and	O
robot	O
soccer	O
as	O
well	O
as	O
conventional	O
games	O
.	O
</s>
<s>
An	O
expert	O
poll	O
around	O
2016	O
,	O
conducted	O
by	O
Katja	O
Grace	O
of	O
the	O
Future	O
of	O
Humanity	O
Institute	O
and	O
associates	O
,	O
gave	O
median	O
estimates	O
of	O
3	O
years	O
for	O
championship	O
Angry	B-Application
Birds	I-Application
,	O
4	O
years	O
for	O
the	O
World	O
Series	O
of	O
Poker	O
,	O
and	O
6	O
years	O
for	O
StarCraft	B-Application
.	O
</s>
<s>
An	O
AI	B-Application
defeated	O
a	O
grandmaster	O
in	O
a	O
regulation	O
tournament	O
game	O
for	O
the	O
first	O
time	O
in	O
1988	O
;	O
rebranded	O
as	O
Deep	B-General_Concept
Blue	I-General_Concept
,	O
it	O
beat	O
the	O
reigning	O
human	O
world	O
chess	B-Application
champion	O
in	O
1997	O
(	O
see	O
Deep	B-General_Concept
Blue	I-General_Concept
versus	O
Garry	O
Kasparov	O
)	O
.	O
</s>
<s>
AlphaGo	B-Application
defeated	O
a	O
European	O
Go	O
champion	O
in	O
October	O
2015	O
,	O
and	O
Lee	O
Sedol	O
in	O
March	O
2016	O
,	O
one	O
of	O
the	O
world	O
's	O
top	O
players	O
(	O
see	O
AlphaGo	B-Application
versus	I-Application
Lee	I-Application
Sedol	I-Application
)	O
.	O
</s>
<s>
AI	B-Application
pioneer	O
and	O
economist	O
Herbert	O
A	O
.	O
Simon	O
inaccurately	O
predicted	O
in	O
1965	O
:	O
"	O
Machines	O
will	O
be	O
capable	O
,	O
within	O
twenty	O
years	O
,	O
of	O
doing	O
any	O
work	O
a	O
man	O
can	O
do	O
"	O
.	O
</s>
<s>
Similarly	O
,	O
in	O
1970	O
Marvin	O
Minsky	O
wrote	O
that	O
"	O
Within	O
a	O
generation	O
...	O
the	O
problem	O
of	O
creating	O
artificial	B-Application
intelligence	I-Application
will	O
substantially	O
be	O
solved.	O
"	O
</s>
<s>
The	O
median	O
response	O
for	O
when	O
"	O
AI	B-Application
researcher	O
"	O
could	O
be	O
fully	O
automated	O
was	O
around	O
90	O
years	O
.	O
</s>
