<s>
AlphaGo	B-Application
is	O
a	O
computer	B-Application
program	I-Application
that	O
plays	O
the	O
board	O
game	O
Go	O
.	O
</s>
<s>
It	O
was	O
developed	O
by	O
the	O
London-based	O
DeepMind	B-Application
Technologies	I-Application
,	O
an	O
acquired	O
subsidiary	O
of	O
Google	B-Application
(	O
now	O
Alphabet	O
Inc	O
.	O
)	O
.	O
</s>
<s>
Subsequent	O
versions	O
of	O
AlphaGo	B-Application
became	O
increasingly	O
powerful	O
,	O
including	O
a	O
version	O
that	O
competed	O
under	O
the	O
name	O
Master	B-Application
.	O
</s>
<s>
After	O
retiring	O
from	O
competitive	O
play	O
,	O
AlphaGo	B-Application
Master	I-Application
was	O
succeeded	O
by	O
an	O
even	O
more	O
powerful	O
version	O
known	O
as	O
AlphaGo	B-Application
Zero	I-Application
,	O
which	O
was	O
completely	O
self-taught	B-Algorithm
without	O
learning	O
from	O
human	O
games	O
.	O
</s>
<s>
AlphaGo	B-Application
Zero	I-Application
was	O
then	O
generalized	O
into	O
a	O
program	O
known	O
as	O
AlphaZero	B-Application
,	O
which	O
played	O
additional	O
games	O
,	O
including	O
chess	B-Application
and	O
shogi	O
.	O
</s>
<s>
AlphaZero	B-Application
has	O
in	O
turn	O
been	O
succeeded	O
by	O
a	O
program	O
known	O
as	O
MuZero	B-Application
which	O
learns	O
without	O
being	O
taught	O
the	O
rules	O
.	O
</s>
<s>
AlphaGo	B-Application
and	O
its	O
successors	O
use	O
a	O
Monte	B-Application
Carlo	I-Application
tree	I-Application
search	I-Application
algorithm	O
to	O
find	O
its	O
moves	O
based	O
on	O
knowledge	O
previously	O
acquired	O
by	O
machine	O
learning	O
,	O
specifically	O
by	O
an	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
(	O
a	O
deep	B-Algorithm
learning	I-Algorithm
method	O
)	O
by	O
extensive	O
training	O
,	O
both	O
from	O
human	O
and	O
computer	O
play	O
.	O
</s>
<s>
A	O
neural	B-Architecture
network	I-Architecture
is	O
trained	O
to	O
identify	O
the	O
best	O
moves	O
and	O
the	O
winning	O
percentages	O
of	O
these	O
moves	O
.	O
</s>
<s>
This	O
neural	B-Architecture
network	I-Architecture
improves	O
the	O
strength	O
of	O
the	O
tree	B-Algorithm
search	I-Algorithm
,	O
resulting	O
in	O
stronger	O
move	O
selection	O
in	O
the	O
next	O
iteration	O
.	O
</s>
<s>
In	O
October	O
2015	O
,	O
in	O
a	O
match	O
against	O
Fan	O
Hui	O
,	O
the	O
original	O
AlphaGo	B-Application
became	O
the	O
first	O
computer	B-Application
Go	I-Application
program	O
to	O
beat	O
a	O
human	O
professional	O
Go	O
player	O
without	O
handicap	O
on	O
a	O
full-sized	O
19×19	O
board	O
.	O
</s>
<s>
In	O
March	O
2016	O
,	O
it	O
beat	O
Lee	O
Sedol	O
in	O
a	B-Application
five-game	I-Application
match	I-Application
,	O
the	O
first	O
time	O
a	O
computer	B-Application
Go	I-Application
program	O
has	O
beaten	O
a	O
9-dan	O
professional	O
without	O
handicap	O
.	O
</s>
<s>
Although	O
it	O
lost	O
to	O
Lee	O
Sedol	O
in	O
the	O
fourth	O
game	O
,	O
Lee	O
resigned	O
in	O
the	O
final	O
game	O
,	O
giving	O
a	O
final	O
score	O
of	O
4	O
games	O
to	O
1	O
in	O
favour	O
of	O
AlphaGo	B-Application
.	O
</s>
<s>
In	O
recognition	O
of	O
the	O
victory	O
,	O
AlphaGo	B-Application
was	O
awarded	O
an	O
honorary	O
9-dan	O
by	O
the	O
Korea	O
Baduk	O
Association	O
.	O
</s>
<s>
The	O
lead	O
up	O
and	O
the	O
challenge	O
match	O
with	O
Lee	O
Sedol	O
were	O
documented	O
in	O
a	O
documentary	O
film	O
also	O
titled	O
AlphaGo	B-Application
,	O
directed	O
by	O
Greg	B-Application
Kohs	I-Application
.	O
</s>
<s>
The	O
win	O
by	O
AlphaGo	B-Application
was	O
chosen	O
by	O
Science	O
as	O
one	O
of	O
the	O
Breakthrough	O
of	O
the	O
Year	O
runners-up	O
on	O
22	O
December	O
2016	O
.	O
</s>
<s>
At	O
the	O
2017	O
Future	B-Application
of	I-Application
Go	I-Application
Summit	I-Application
,	O
the	O
Master	B-Application
version	O
of	O
AlphaGo	B-Application
beat	O
Ke	O
Jie	O
,	O
the	O
number	O
one	O
ranked	O
player	O
in	O
the	O
world	O
at	O
the	O
time	O
,	O
in	O
a	O
three-game	B-Application
match	I-Application
,	O
after	O
which	O
AlphaGo	B-Application
was	O
awarded	O
professional	O
9-dan	O
by	O
the	O
Chinese	O
Weiqi	O
Association	O
.	O
</s>
<s>
After	O
the	O
match	O
between	O
AlphaGo	B-Application
and	O
Ke	O
Jie	O
,	O
DeepMind	B-Application
retired	O
AlphaGo	B-Application
,	O
while	O
continuing	O
AI	B-Application
research	I-Application
in	O
other	O
areas	O
.	O
</s>
<s>
The	O
self-taught	B-Algorithm
AlphaGo	B-Application
Zero	I-Application
achieved	O
a	O
100	O
–	O
0	O
victory	O
against	O
the	O
early	O
competitive	O
version	O
of	O
AlphaGo	B-Application
,	O
and	O
its	O
successor	O
AlphaZero	B-Application
is	O
currently	O
perceived	O
as	O
the	O
world	O
's	O
top	O
player	O
in	O
Go	O
.	O
</s>
<s>
Go	O
is	O
considered	O
much	O
more	O
difficult	O
for	O
computers	O
to	O
win	O
than	O
other	O
games	O
such	O
as	O
chess	B-Application
,	O
because	O
its	O
strategic	O
and	O
aesthetic	O
nature	O
makes	O
it	O
hard	O
to	O
directly	O
construct	O
an	O
evaluation	O
function	O
,	O
and	O
its	O
much	O
larger	O
branching	B-Data_Structure
factor	I-Data_Structure
makes	O
it	O
prohibitively	O
difficult	O
to	O
use	O
traditional	O
AI	B-Application
methods	O
such	O
as	O
alpha	B-Algorithm
–	I-Algorithm
beta	I-Algorithm
pruning	I-Algorithm
,	O
tree	B-Algorithm
traversal	I-Algorithm
and	O
heuristic	B-Algorithm
search	O
.	O
</s>
<s>
Almost	O
two	O
decades	O
after	O
IBM	O
's	O
computer	O
Deep	B-General_Concept
Blue	I-General_Concept
beat	O
world	O
chess	B-Application
champion	O
Garry	O
Kasparov	O
in	O
the	O
1997	O
match	O
,	O
the	O
strongest	O
Go	B-Application
programs	I-Application
using	O
artificial	B-Application
intelligence	I-Application
techniques	O
only	O
reached	O
about	O
amateur	O
5-dan	O
level	O
,	O
and	O
still	O
could	O
not	O
beat	O
a	O
professional	O
Go	O
player	O
without	O
a	O
handicap	O
.	O
</s>
<s>
In	O
2012	O
,	O
the	O
software	B-Application
program	I-Application
Zen	B-Application
,	O
running	O
on	O
a	O
four	O
PC	O
cluster	O
,	O
beat	O
Masaki	O
Takemiya	O
(	O
9p	O
)	O
twice	O
at	O
five	O
-	O
and	O
four-stone	O
handicaps	O
.	O
</s>
<s>
In	O
2013	O
,	O
Crazy	B-Application
Stone	I-Application
beat	O
Yoshio	O
Ishida	O
(	O
9p	O
)	O
at	O
a	O
four-stone	O
handicap	O
.	O
</s>
<s>
According	O
to	O
DeepMind	B-Application
's	O
David	O
Silver	O
,	O
the	O
AlphaGo	B-Application
research	O
project	O
was	O
formed	O
around	O
2014	O
to	O
test	O
how	O
well	O
a	O
neural	B-Architecture
network	I-Architecture
using	O
deep	B-Algorithm
learning	I-Algorithm
can	O
compete	O
at	O
Go	O
.	O
</s>
<s>
AlphaGo	B-Application
represents	O
a	O
significant	O
improvement	O
over	O
previous	O
Go	B-Application
programs	I-Application
.	O
</s>
<s>
In	O
500	O
games	O
against	O
other	O
available	O
Go	B-Application
programs	I-Application
,	O
including	O
Crazy	B-Application
Stone	I-Application
and	O
Zen	B-Application
,	O
AlphaGo	B-Application
running	O
on	O
a	O
single	O
computer	O
won	O
all	O
but	O
one	O
.	O
</s>
<s>
In	O
a	O
similar	O
matchup	O
,	O
AlphaGo	B-Application
running	O
on	O
multiple	O
computers	O
won	O
all	O
500	O
games	O
played	O
against	O
other	O
Go	B-Application
programs	I-Application
,	O
and	O
77%	O
of	O
games	O
played	O
against	O
AlphaGo	B-Application
running	O
on	O
a	O
single	O
computer	O
.	O
</s>
<s>
The	O
distributed	B-Architecture
version	O
in	O
October	O
2015	O
was	O
using	O
1,202	O
CPUs	B-General_Concept
and	O
176	O
GPUs	B-Architecture
.	O
</s>
<s>
In	O
October	O
2015	O
,	O
the	O
distributed	B-Architecture
version	O
of	O
AlphaGo	B-Application
defeated	O
the	O
European	O
Go	O
champion	O
Fan	O
Hui	O
,	O
a	O
2-dan	O
(	O
out	O
of	O
9	O
dan	O
possible	O
)	O
professional	O
,	O
five	O
to	O
zero	O
.	O
</s>
<s>
This	O
was	O
the	O
first	O
time	O
a	O
computer	B-Application
Go	I-Application
program	O
had	O
beaten	O
a	O
professional	O
human	O
player	O
on	O
a	O
full-sized	O
board	O
without	O
handicap	O
.	O
</s>
<s>
AlphaGo	B-Application
played	O
South	O
Korean	O
professional	O
Go	O
player	O
Lee	O
Sedol	O
,	O
ranked	O
9-dan	O
,	O
one	O
of	O
the	O
best	O
players	O
at	O
Go	O
,	O
with	O
five	O
games	O
taking	O
place	O
at	O
the	O
Four	O
Seasons	O
Hotel	O
in	O
Seoul	O
,	O
South	O
Korea	O
on	O
9	O
,	O
10	O
,	O
12	O
,	O
13	O
,	O
and	O
15	O
March	O
2016	O
,	O
which	O
were	O
video-streamed	O
live	O
.	O
</s>
<s>
Out	O
of	O
five	O
games	O
,	O
AlphaGo	B-Application
won	O
four	O
games	O
and	O
Lee	O
won	O
the	O
fourth	O
game	O
which	O
made	O
him	O
recorded	O
as	O
the	O
only	O
human	O
player	O
who	O
beat	O
AlphaGo	B-Application
in	O
all	O
of	O
its	O
74	O
official	O
games	O
.	O
</s>
<s>
AlphaGo	B-Application
ran	O
on	O
Google	B-Application
's	I-Application
cloud	O
computing	O
with	O
its	O
servers	O
located	O
in	O
the	O
United	O
States	O
.	O
</s>
<s>
The	O
version	O
of	O
AlphaGo	B-Application
playing	O
against	O
Lee	O
used	O
a	O
similar	O
amount	O
of	O
computing	O
power	O
as	O
was	O
used	O
in	O
the	O
Fan	O
Hui	O
match	O
.	O
</s>
<s>
The	O
Economist	O
reported	O
that	O
it	O
used	O
1,920	O
CPUs	B-General_Concept
and	O
280	O
GPUs	B-Architecture
.	O
</s>
<s>
AlphaGo	B-Application
was	O
not	O
specifically	O
trained	O
to	O
face	O
Lee	O
nor	O
was	O
designed	O
to	O
compete	O
with	O
any	O
specific	O
human	O
players	O
.	O
</s>
<s>
The	O
first	O
three	O
games	O
were	O
won	O
by	O
AlphaGo	B-Application
following	O
resignations	O
by	O
Lee	O
.	O
</s>
<s>
However	O
,	O
Lee	O
beat	O
AlphaGo	B-Application
in	O
the	O
fourth	O
game	O
,	O
winning	O
by	O
resignation	O
at	O
move	O
180	O
.	O
</s>
<s>
AlphaGo	B-Application
then	O
continued	O
to	O
achieve	O
a	O
fourth	O
win	O
,	O
winning	O
the	O
fifth	O
game	O
by	O
resignation	O
.	O
</s>
<s>
Since	O
AlphaGo	B-Application
won	O
four	O
out	O
of	O
five	O
and	O
thus	O
the	O
series	O
,	O
the	O
prize	O
will	O
be	O
donated	O
to	O
charities	O
,	O
including	O
UNICEF	O
.	O
</s>
<s>
In	O
June	O
2016	O
,	O
at	O
a	O
presentation	O
held	O
at	O
a	O
university	O
in	O
the	O
Netherlands	O
,	O
Aja	O
Huang	O
,	O
one	O
of	O
the	O
Deep	O
Mind	O
team	O
,	O
revealed	O
that	O
they	O
had	O
patched	O
the	O
logical	O
weakness	O
that	O
occurred	O
during	O
the	O
4th	O
game	O
of	O
the	O
match	O
between	O
AlphaGo	B-Application
and	O
Lee	O
,	O
and	O
that	O
after	O
move	O
78	O
(	O
which	O
was	O
dubbed	O
the	O
"	O
divine	O
move	O
"	O
by	O
many	O
professionals	O
)	O
,	O
it	O
would	O
play	O
as	O
intended	O
and	O
maintain	O
Black	O
's	O
advantage	O
.	O
</s>
<s>
Before	O
move	O
78	O
,	O
AlphaGo	B-Application
was	O
leading	O
throughout	O
the	O
game	O
,	O
but	O
Lee	O
's	O
move	O
caused	O
the	O
program	O
's	O
computing	O
powers	O
to	O
be	O
diverted	O
and	O
confused	O
.	O
</s>
<s>
Huang	O
explained	O
that	O
AlphaGo	B-Application
's	O
policy	O
network	O
of	O
finding	O
the	O
most	O
accurate	O
move	O
order	O
and	O
continuation	O
did	O
not	O
precisely	O
guide	O
AlphaGo	B-Application
to	O
make	O
the	O
correct	O
continuation	O
after	O
move	O
78	O
,	O
since	O
its	O
value	O
network	O
did	O
not	O
determine	O
Lee	O
's	O
78th	O
move	O
as	O
being	O
the	O
most	O
likely	O
,	O
and	O
therefore	O
when	O
the	O
move	O
was	O
made	O
AlphaGo	B-Application
could	O
not	O
make	O
the	O
right	O
adjustment	O
to	O
the	O
logical	O
continuation	O
.	O
</s>
<s>
On	O
29	O
December	O
2016	O
,	O
a	O
new	O
account	O
on	O
the	O
Tygem	B-Application
server	O
named	O
"	O
Magister	O
"	O
(	O
shown	O
as	O
'	O
Magist	B-Application
 '	O
at	O
the	O
server	O
's	O
Chinese	O
version	O
)	O
from	O
South	O
Korea	O
began	O
to	O
play	O
games	O
with	O
professional	O
players	O
.	O
</s>
<s>
It	O
changed	O
its	O
account	O
name	O
to	O
"	O
Master	B-Application
"	O
on	O
30	O
December	O
,	O
then	O
moved	O
to	O
the	O
FoxGo	O
server	O
on	O
1	O
January	O
2017	O
.	O
</s>
<s>
On	O
4	O
January	O
,	O
DeepMind	B-Application
confirmed	O
that	O
the	O
"	O
Magister	O
"	O
and	O
the	O
"	O
Master	B-Application
"	O
were	O
both	O
played	O
by	O
an	O
updated	O
version	O
of	O
AlphaGo	B-Application
,	O
called	O
AlphaGo	B-Application
Master	I-Application
.	O
</s>
<s>
As	O
of	O
5	O
January	O
2017	O
,	O
AlphaGo	B-Application
Master	I-Application
's	O
online	O
record	O
was	O
60	O
wins	O
and	O
0	O
losses	O
,	O
including	O
three	O
victories	O
over	O
Go	O
's	O
top-ranked	O
player	O
,	O
Ke	O
Jie	O
,	O
who	O
had	O
been	O
quietly	O
briefed	O
in	O
advance	O
that	O
Master	B-Application
was	O
a	O
version	O
of	O
AlphaGo	B-Application
.	O
</s>
<s>
After	O
losing	O
to	O
Master	B-Application
,	O
Gu	O
Li	O
offered	O
a	O
bounty	O
of	O
100,000	O
yuan	O
(	O
US$	O
14,400	O
)	O
to	O
the	O
first	O
human	O
player	O
who	O
could	O
defeat	O
Master	B-Application
.	O
</s>
<s>
Master	B-Application
played	O
at	O
the	O
pace	O
of	O
10	O
games	O
per	O
day	O
.	O
</s>
<s>
Many	O
quickly	O
suspected	O
it	O
to	O
be	O
an	O
AI	B-Application
player	O
due	O
to	O
little	O
or	O
no	O
resting	O
between	O
games	O
.	O
</s>
<s>
Master	B-Application
offered	O
to	O
extend	O
the	O
byo-yomi	O
to	O
one	O
minute	O
when	O
playing	O
with	O
Nie	O
Weiping	O
in	O
consideration	O
of	O
his	O
age	O
.	O
</s>
<s>
After	O
winning	O
its	O
59th	O
game	O
Master	B-Application
revealed	O
itself	O
in	O
the	O
chatroom	O
to	O
be	O
controlled	O
by	O
Dr.	O
Aja	O
Huang	O
of	O
the	O
DeepMind	B-Application
team	O
,	O
then	O
changed	O
its	O
nationality	O
to	O
the	O
United	O
Kingdom	O
.	O
</s>
<s>
After	O
these	O
games	O
were	O
completed	O
,	O
the	O
co-founder	O
of	O
DeepMind	B-Application
,	O
Demis	O
Hassabis	O
,	O
said	O
in	O
a	O
tweet	O
,	O
"	O
we	O
're	O
looking	O
forward	O
to	O
playing	O
some	O
official	O
,	O
full-length	O
games	O
later	O
 [ 2017 ] 	O
in	O
collaboration	O
with	O
Go	O
organizations	O
and	O
experts	O
"	O
.	O
</s>
<s>
In	O
the	O
Future	B-Application
of	I-Application
Go	I-Application
Summit	I-Application
held	O
in	O
Wuzhen	O
in	O
May	O
2017	O
,	O
AlphaGo	B-Application
Master	I-Application
played	O
three	O
games	O
with	O
Ke	O
Jie	O
,	O
the	O
world	O
No.1	O
ranked	O
player	O
,	O
as	O
well	O
as	O
two	O
games	O
with	O
several	O
top	O
Chinese	O
professionals	O
,	O
one	O
pair	O
Go	O
game	O
and	O
one	O
against	O
a	O
collaborating	O
team	O
of	O
five	O
human	O
players	O
.	O
</s>
<s>
Google	B-Application
DeepMind	I-Application
offered	O
1.5	O
million	O
dollar	O
winner	O
prizes	O
for	O
the	O
three-game	B-Application
match	I-Application
between	O
Ke	O
Jie	O
and	O
Master	B-Application
while	O
the	O
losing	O
side	O
took	O
300,000	O
dollars	O
.	O
</s>
<s>
Master	B-Application
won	O
all	O
three	O
games	O
against	O
Ke	O
Jie	O
,	O
after	O
which	O
AlphaGo	B-Application
was	O
awarded	O
professional	O
9-dan	O
by	O
the	O
Chinese	O
Weiqi	O
Association	O
.	O
</s>
<s>
After	O
winning	O
its	O
three-game	B-Application
match	I-Application
against	O
Ke	O
Jie	O
,	O
the	O
top-rated	O
world	O
Go	O
player	O
,	O
AlphaGo	B-Application
retired	O
.	O
</s>
<s>
DeepMind	B-Application
also	O
disbanded	O
the	O
team	O
that	O
worked	O
on	O
the	O
game	O
to	O
focus	O
on	O
AI	B-Application
research	I-Application
in	O
other	O
areas	O
.	O
</s>
<s>
After	O
the	O
Summit	O
,	O
Deepmind	B-Application
published	O
50	O
full	O
length	O
AlphaGo	B-Application
vs	O
AlphaGo	B-Application
matches	O
,	O
as	O
a	O
gift	O
to	O
the	O
Go	O
community	O
.	O
</s>
<s>
AlphaGo	B-Application
's	O
team	O
published	O
an	O
article	O
in	O
the	O
journal	O
Nature	O
on	O
19	O
October	O
2017	O
,	O
introducing	O
AlphaGo	B-Application
Zero	I-Application
,	O
a	O
version	O
without	O
human	O
data	O
and	O
stronger	O
than	O
any	O
previous	O
human-champion-defeating	O
version	O
.	O
</s>
<s>
By	O
playing	O
games	O
against	O
itself	O
,	O
AlphaGo	B-Application
Zero	I-Application
surpassed	O
the	O
strength	O
of	O
AlphaGo	B-Application
Lee	I-Application
in	O
three	O
days	O
by	O
winning	O
100	O
games	O
to	O
0	O
,	O
reached	O
the	O
level	O
of	O
AlphaGo	B-Application
Master	I-Application
in	O
21	O
days	O
,	O
and	O
exceeded	O
all	O
the	O
old	O
versions	O
in	O
40	O
days	O
.	O
</s>
<s>
In	O
a	O
paper	O
released	O
on	O
arXiv	O
on	O
5	O
December	O
2017	O
,	O
DeepMind	B-Application
claimed	O
that	O
it	O
generalized	O
AlphaGo	B-Application
Zero	I-Application
's	O
approach	O
into	O
a	O
single	O
AlphaZero	B-Application
algorithm	O
,	O
which	O
achieved	O
within	O
24	O
hours	O
a	O
superhuman	O
level	O
of	O
play	O
in	O
the	O
games	O
of	O
chess	B-Application
,	O
shogi	O
,	O
and	O
Go	O
by	O
defeating	O
world-champion	O
programs	O
,	O
Stockfish	B-Application
,	O
Elmo	B-Application
,	O
and	O
3-day	O
version	O
of	O
AlphaGo	B-Application
Zero	I-Application
in	O
each	O
case	O
.	O
</s>
<s>
On	O
11	O
December	O
2017	O
,	O
DeepMind	B-Application
released	O
AlphaGo	B-Application
teaching	O
tool	O
on	O
its	O
website	O
to	O
analyze	O
winning	O
rates	O
of	O
different	O
Go	O
openings	O
as	O
calculated	O
by	O
AlphaGo	B-Application
Master	I-Application
.	O
</s>
<s>
The	O
teaching	O
tool	O
collects	O
6,000	O
Go	O
openings	O
from	O
230,000	O
human	O
games	O
each	O
analyzed	O
with	O
10,000,000	O
simulations	O
by	O
AlphaGo	B-Application
Master	I-Application
.	O
</s>
<s>
An	O
early	O
version	O
of	O
AlphaGo	B-Application
was	O
tested	O
on	O
hardware	O
with	O
various	O
numbers	O
of	O
CPUs	B-General_Concept
and	O
GPUs	B-Architecture
,	O
running	O
in	O
asynchronous	O
or	O
distributed	B-Architecture
mode	O
.	O
</s>
<s>
In	O
May	O
2016	O
,	O
Google	B-Application
unveiled	O
its	O
own	O
proprietary	O
hardware	O
"	O
tensor	B-Device
processing	I-Device
units	I-Device
"	O
,	O
which	O
it	O
stated	O
had	O
already	O
been	O
deployed	O
in	O
multiple	O
internal	O
projects	O
at	O
Google	B-Application
,	O
including	O
the	O
AlphaGo	B-Application
match	O
against	O
Lee	O
Sedol	O
.	O
</s>
<s>
In	O
the	O
Future	B-Application
of	I-Application
Go	I-Application
Summit	I-Application
in	O
May	O
2017	O
,	O
DeepMind	B-Application
disclosed	O
that	O
the	O
version	O
of	O
AlphaGo	B-Application
used	O
in	O
this	O
Summit	O
was	O
AlphaGo	B-Application
Master	I-Application
,	O
and	O
revealed	O
that	O
it	O
had	O
measured	O
the	O
strength	O
of	O
different	O
versions	O
of	O
the	O
software	O
.	O
</s>
<s>
AlphaGo	B-Application
Lee	I-Application
,	O
the	O
version	O
used	O
against	O
Lee	O
,	O
could	O
give	O
AlphaGo	B-Application
Fan	I-Application
,	O
the	O
version	O
used	O
in	O
AlphaGo	B-Application
vs	O
.	O
</s>
<s>
Fan	O
Hui	O
,	O
three	O
stones	O
,	O
and	O
AlphaGo	B-Application
Master	I-Application
was	O
even	O
three	O
stones	O
stronger	O
.	O
</s>
<s>
As	O
of	O
2016	O
,	O
AlphaGo	B-Application
's	O
algorithm	O
uses	O
a	O
combination	O
of	O
machine	O
learning	O
and	O
tree	B-Algorithm
search	I-Algorithm
techniques	O
,	O
combined	O
with	O
extensive	O
training	O
,	O
both	O
from	O
human	O
and	O
computer	O
play	O
.	O
</s>
<s>
It	O
uses	O
Monte	B-Application
Carlo	I-Application
tree	I-Application
search	I-Application
,	O
guided	O
by	O
a	O
"	O
value	O
network	O
"	O
and	O
a	O
"	O
policy	O
network	O
,	O
"	O
both	O
implemented	O
using	O
deep	O
neural	B-Architecture
network	I-Architecture
technology	O
.	O
</s>
<s>
A	O
limited	O
amount	O
of	O
game-specific	O
feature	O
detection	O
pre-processing	O
(	O
for	O
example	O
,	O
to	O
highlight	O
whether	O
a	O
move	O
matches	O
a	O
nakade	O
pattern	O
)	O
is	O
applied	O
to	O
the	O
input	O
before	O
it	O
is	O
sent	O
to	O
the	O
neural	B-Architecture
networks	I-Architecture
.	O
</s>
<s>
The	O
networks	O
are	O
convolutional	B-Architecture
neural	I-Architecture
networks	I-Architecture
with	O
12	O
layers	O
,	O
trained	O
by	O
reinforcement	O
learning	O
.	O
</s>
<s>
The	O
system	O
's	O
neural	B-Architecture
networks	I-Architecture
were	O
initially	O
bootstrapped	O
from	O
human	O
gameplay	O
expertise	O
.	O
</s>
<s>
AlphaGo	B-Application
was	O
initially	O
trained	O
to	O
mimic	O
human	O
play	O
by	O
attempting	O
to	O
match	O
the	O
moves	O
of	O
expert	O
players	O
from	O
recorded	O
historical	O
games	O
,	O
using	O
a	O
database	O
of	O
around	O
30	O
million	O
moves	O
.	O
</s>
<s>
Toby	O
Manning	O
,	O
the	O
match	O
referee	O
for	O
AlphaGo	B-Application
vs	O
.	O
</s>
<s>
AlphaGo	B-Application
's	O
playing	O
style	O
strongly	O
favours	O
greater	O
probability	O
of	O
winning	O
by	O
fewer	O
points	O
over	O
lesser	O
probability	O
of	O
winning	O
by	O
more	O
points	O
.	O
</s>
<s>
AlphaGo	B-Application
's	O
March	O
2016	O
victory	O
was	O
a	O
major	O
milestone	O
in	O
artificial	B-Application
intelligence	I-Application
research	I-Application
.	O
</s>
<s>
Most	O
experts	O
thought	O
a	O
Go	O
program	O
as	O
powerful	O
as	O
AlphaGo	B-Application
was	O
at	O
least	O
five	O
years	O
away	O
;	O
some	O
experts	O
thought	O
that	O
it	O
would	O
take	O
at	O
least	O
another	O
decade	O
before	O
computers	O
would	O
beat	O
Go	O
champions	O
.	O
</s>
<s>
Most	O
observers	O
at	O
the	O
beginning	O
of	O
the	O
2016	O
matches	O
expected	O
Lee	O
to	O
beat	O
AlphaGo	B-Application
.	O
</s>
<s>
With	O
games	O
such	O
as	O
checkers	O
(	O
that	O
has	O
been	O
"	O
solved	O
"	O
by	O
the	O
Chinook	B-General_Concept
draughts	I-General_Concept
player	I-General_Concept
team	O
)	O
,	O
chess	B-Application
,	O
and	O
now	O
Go	O
won	O
by	O
computers	O
,	O
victories	O
at	O
popular	O
board	O
games	O
can	O
no	O
longer	O
serve	O
as	O
major	O
milestones	O
for	O
artificial	B-Application
intelligence	I-Application
in	O
the	O
way	O
that	O
they	O
used	O
to	O
.	O
</s>
<s>
Deep	B-General_Concept
Blue	I-General_Concept
's	O
Murray	O
Campbell	O
called	O
AlphaGo	B-Application
's	O
victory	O
"	O
the	O
end	O
of	O
an	O
era	O
...	O
board	O
games	O
are	O
more	O
or	O
less	O
done	O
and	O
it	O
's	O
time	O
to	O
move	O
on.	O
"	O
</s>
<s>
When	O
compared	O
with	O
Deep	B-General_Concept
Blue	I-General_Concept
or	O
Watson	B-Application
,	O
AlphaGo	B-Application
's	O
underlying	O
algorithms	O
are	O
potentially	O
more	O
general-purpose	O
and	O
may	O
be	O
evidence	O
that	O
the	O
scientific	O
community	O
is	O
making	O
progress	O
towards	O
artificial	O
general	O
intelligence	O
.	O
</s>
<s>
Some	O
commentators	O
believe	O
AlphaGo	B-Application
's	O
victory	O
makes	O
for	O
a	O
good	O
opportunity	O
for	O
society	O
to	O
start	O
preparing	O
for	O
the	O
possible	O
future	O
impact	O
of	O
machines	O
with	O
general	O
purpose	O
intelligence	O
.	O
</s>
<s>
As	O
noted	O
by	O
entrepreneur	O
Guy	O
Suter	O
,	O
AlphaGo	B-Application
only	O
knows	O
how	O
to	O
play	O
Go	O
and	O
does	O
n't	O
possess	O
general-purpose	O
intelligence	O
;	O
"[It]	O
could	O
n't	O
just	O
wake	O
up	O
one	O
morning	O
and	O
decide	O
it	O
wants	O
to	O
learn	O
how	O
to	O
use	O
firearms.	O
"	O
</s>
<s>
AI	B-Application
researcher	O
Stuart	O
Russell	O
said	O
that	O
AI	B-Application
systems	O
such	O
as	O
AlphaGo	B-Application
have	O
progressed	O
quicker	O
and	O
become	O
more	O
powerful	O
than	O
expected	O
,	O
and	O
we	O
must	O
therefore	O
develop	O
methods	O
to	O
ensure	O
they	O
"	O
remain	O
under	O
human	O
control	O
"	O
.	O
</s>
<s>
Some	O
scholars	O
,	O
such	O
as	O
Stephen	O
Hawking	O
,	O
warned	O
(	O
in	O
May	O
2015	O
before	O
the	O
matches	O
)	O
that	O
some	O
future	O
self-improving	O
AI	B-Application
could	O
gain	O
actual	O
general	O
intelligence	O
,	O
leading	O
to	O
an	O
unexpected	O
AI	B-Application
takeover	O
;	O
other	O
scholars	O
disagree	O
:	O
AI	B-Application
expert	O
Jean-Gabriel	O
Ganascia	O
believes	O
that	O
"	O
Things	O
like	O
'	O
common	O
sense	O
 '	O
...	O
may	O
never	O
be	O
reproducible	O
"	O
,	O
and	O
says	O
"	O
I	O
do	O
n't	O
see	O
why	O
we	O
would	O
speak	O
about	O
fears	O
.	O
</s>
<s>
In	O
China	O
,	O
AlphaGo	B-Application
was	O
a	O
"	O
Sputnik	O
moment	O
"	O
which	O
helped	O
convince	O
the	O
Chinese	O
government	O
to	O
prioritize	O
and	O
dramatically	O
increase	O
funding	O
for	O
artificial	B-Application
intelligence	I-Application
.	I-Application
</s>
<s>
In	O
2017	O
,	O
the	O
DeepMind	B-Application
AlphaGo	I-Application
team	O
received	O
the	O
inaugural	O
IJCAI	O
Marvin	O
Minsky	O
medal	O
for	O
Outstanding	O
Achievements	O
in	O
AI	B-Application
.	O
</s>
<s>
"	O
AlphaGo	B-Application
is	O
a	O
wonderful	O
achievement	O
,	O
and	O
a	O
perfect	O
example	O
of	O
what	O
the	O
Minsky	O
Medal	O
was	O
initiated	O
to	O
recognise	O
"	O
,	O
said	O
Professor	O
Michael	O
Wooldridge	O
,	O
Chair	O
of	O
the	O
IJCAI	O
Awards	O
Committee	O
.	O
</s>
<s>
"	O
What	O
particularly	O
impressed	O
IJCAI	O
was	O
that	O
AlphaGo	B-Application
achieves	O
what	O
it	O
does	O
through	O
a	O
brilliant	O
combination	O
of	O
classic	O
AI	B-Application
techniques	O
as	O
well	O
as	O
the	O
state-of-the-art	O
machine	O
learning	O
techniques	O
that	O
DeepMind	B-Application
is	O
so	O
closely	O
associated	O
with	O
.	O
</s>
<s>
It	O
's	O
a	O
breathtaking	O
demonstration	O
of	O
contemporary	O
AI	B-Application
,	O
and	O
we	O
are	O
delighted	O
to	O
be	O
able	O
to	O
recognise	O
it	O
with	O
this	O
award.	O
"	O
</s>
<s>
Many	O
top	O
Go	O
players	O
characterized	O
AlphaGo	B-Application
's	O
unorthodox	O
plays	O
as	O
seemingly-questionable	O
moves	O
that	O
initially	O
befuddled	O
onlookers	O
,	O
but	O
made	O
sense	O
in	O
hindsight	O
:	O
"	O
All	O
but	O
the	O
very	O
best	O
Go	O
players	O
craft	O
their	O
style	O
by	O
imitating	O
top	O
players	O
.	O
</s>
<s>
AlphaGo	B-Application
seems	O
to	O
have	O
totally	O
original	O
moves	O
it	O
creates	O
itself.	O
"	O
</s>
<s>
AlphaGo	B-Application
appeared	O
to	O
have	O
unexpectedly	O
become	O
much	O
stronger	O
,	O
even	O
when	O
compared	O
with	O
its	O
October	O
2015	O
match	O
where	O
a	O
computer	O
had	O
beaten	O
a	O
Go	O
professional	O
for	O
the	O
first	O
time	O
ever	O
without	O
the	O
advantage	O
of	O
a	O
handicap	O
.	O
</s>
<s>
The	O
Korea	O
Baduk	O
Association	O
,	O
the	O
organization	O
that	O
oversees	O
Go	O
professionals	O
in	O
South	O
Korea	O
,	O
awarded	O
AlphaGo	B-Application
an	O
honorary	O
9-dan	O
title	O
for	O
exhibiting	O
creative	O
skills	O
and	O
pushing	O
forward	O
the	O
game	O
's	O
progress	O
.	O
</s>
<s>
China	O
's	O
Ke	O
Jie	O
,	O
an	O
18-year-old	O
generally	O
recognized	O
as	O
the	O
world	O
's	O
best	O
Go	O
player	O
at	O
the	O
time	O
,	O
initially	O
claimed	O
that	O
he	O
would	O
be	O
able	O
to	O
beat	O
AlphaGo	B-Application
,	O
but	O
declined	O
to	O
play	O
against	O
it	O
for	O
fear	O
that	O
it	O
would	O
"	O
copy	O
my	O
style	O
"	O
.	O
</s>
<s>
As	O
the	O
matches	O
progressed	O
,	O
Ke	O
Jie	O
went	O
back	O
and	O
forth	O
,	O
stating	O
that	O
"	O
it	O
is	O
highly	O
likely	O
that	O
I	O
(	O
could	O
)	O
lose	O
"	O
after	O
analysing	O
the	O
first	O
three	O
matches	O
,	O
but	O
regaining	O
confidence	O
after	O
AlphaGo	B-Application
displayed	O
flaws	O
in	O
the	O
fourth	O
match	O
.	O
</s>
<s>
Toby	O
Manning	O
,	O
the	O
referee	O
of	O
AlphaGo	B-Application
's	O
match	O
against	O
Fan	O
Hui	O
,	O
and	O
Hajin	O
Lee	O
,	O
secretary	O
general	O
of	O
the	O
International	O
Go	O
Federation	O
,	O
both	O
reason	O
that	O
in	O
the	O
future	O
,	O
Go	O
players	O
will	O
get	O
help	O
from	O
computers	O
to	O
learn	O
what	O
they	O
have	O
done	O
wrong	O
in	O
games	O
and	O
improve	O
their	O
skills	O
.	O
</s>
<s>
It	O
was	O
AlphaGo	B-Application
's	O
total	O
victory.	O
"	O
</s>
<s>
Lee	O
apologized	O
for	O
his	O
losses	O
,	O
stating	O
after	O
game	O
three	O
that	O
"	O
I	O
misjudged	O
the	O
capabilities	O
of	O
AlphaGo	B-Application
and	O
felt	O
powerless.	O
"	O
</s>
<s>
There	O
is	O
a	O
large	O
interest	O
prior	O
to	O
the	O
match	O
between	O
Lee	O
Sedol	O
and	O
AlphaGo	B-Application
in	O
2016	O
.	O
</s>
<s>
Hence	O
,	O
the	O
director	O
Greg	B-Application
Kohs	I-Application
brings	O
in	O
the	O
idea	O
of	O
creating	O
a	O
documentary	O
film	O
to	O
record	O
this	O
historical	O
moment	O
for	O
the	O
public	O
view	O
.	O
</s>
<s>
For	O
example	O
,	O
the	O
close-up	O
shots	O
of	O
Lee	O
Sedol	O
when	O
he	O
realizes	O
that	O
the	O
AlphaGo	B-Application
AI	B-Application
is	O
intelligent	O
,	O
the	O
atmospheric	O
scene	O
of	O
the	O
Korean	O
commentator	O
's	O
distress	O
and	O
affliction	O
following	O
the	O
first	O
defeat	O
,	O
and	O
the	O
tension	O
being	O
held	O
inside	O
the	O
room	O
.	O
</s>
<s>
The	O
documentary	O
also	O
tells	O
a	O
story	O
by	O
describing	O
the	O
background	O
of	O
AlphaGo	B-Application
technology	O
and	O
the	O
customs	O
of	O
the	O
Korean	O
Go	O
community	O
.	O
</s>
<s>
For	O
instance	O
,	O
the	O
details	O
of	O
the	O
AI	B-Application
prior	O
to	O
AlphaGo	B-Application
,	O
the	O
confidence	O
and	O
pride	O
of	O
the	O
professional	O
Go	O
players	O
,	O
and	O
the	O
shifting	O
of	O
perspective	O
to	O
the	O
Go	O
AI	B-Application
between	O
and	O
after	O
the	O
match	O
as	O
“	O
If	O
anything	O
could	O
be	O
added	O
,	O
I	O
would	O
include	O
information	O
about	O
the	O
primitive	O
level	O
of	O
top	O
Go	O
A.I.s	O
before	O
AlphaGo	B-Application
,	O
and	O
more	O
about	O
professional	O
Go	O
players’	O
lives	O
and	O
pride	O
,	O
to	O
provide	O
more	O
context	O
for	O
Lee	O
Sedol	O
’s	O
pre-match	O
confidence	O
,	O
and	O
Go	O
players’	O
changing	O
perception	O
of	O
AlphaGo	B-Application
as	O
the	O
match	O
advanced	O
”	O
.	O
</s>
<s>
Fan	O
Hui	O
,	O
a	O
professional	O
Go	O
player	O
,	O
and	O
former	O
player	O
with	O
AlphaGo	B-Application
said	O
that	O
“	O
DeepMind	B-Application
had	O
trained	O
AlphaGo	B-Application
by	O
showing	O
it	O
many	O
strong	O
amateur	O
games	O
of	O
Go	O
to	O
develop	O
its	O
understanding	O
of	O
how	O
a	O
human	O
plays	O
before	O
challenging	O
it	O
to	O
play	O
versions	O
of	O
itself	O
thousands	O
of	O
times	O
,	O
a	O
novel	O
form	O
of	O
reinforcement	O
learning	O
which	O
had	O
given	O
it	O
the	O
ability	O
to	O
rival	O
an	O
expert	O
human	O
.	O
</s>
<s>
The	O
AlphaGo	B-Application
footage	O
slowly	O
captures	O
the	O
moment	O
when	O
Lee	O
Sedol	O
acknowledges	O
the	O
true	O
power	O
of	O
AlphaGo	B-Application
AI	B-Application
.	O
</s>
<s>
In	O
the	O
first	O
game	O
,	O
he	O
had	O
more	O
experience	O
than	O
his	O
human-programmed	O
AI	B-Application
,	O
so	O
he	O
thought	O
it	O
would	O
be	O
easy	O
to	O
beat	O
the	O
AI	B-Application
.	O
</s>
<s>
Also	O
,	O
he	O
spends	O
12	O
minutes	O
on	O
one	O
move	O
,	O
while	O
AlphaGo	B-Application
only	O
takes	O
a	O
minute	O
and	O
a	O
half	O
to	O
respond	O
.	O
</s>
<s>
AlphaGo	B-Application
weighs	O
each	O
alternative	O
equally	O
and	O
consistently	O
.	O
</s>
<s>
Dealing	O
with	O
the	O
problems	O
of	O
AI	B-Application
will	O
take	O
a	O
perspective	O
that	O
looks	O
beyond	O
individual	O
battles	O
.	O
</s>
<s>
Murray	O
Shanahan	O
,	O
a	O
professor	O
of	O
cognitive	O
robotics	O
at	O
Imperial	O
College	O
London	O
,	O
critics	O
that	O
“	O
Go	O
is	O
an	O
extraordinary	O
game	O
but	O
it	O
represents	O
what	O
we	O
can	O
do	O
with	O
AI	B-Application
in	O
all	O
kinds	O
of	O
other	O
spheres	O
,	O
”	O
says	O
Murray	O
Shanahan	O
,	O
professor	O
of	O
cognitive	O
robotics	O
at	O
Imperial	O
College	O
London	O
and	O
senior	O
research	O
scientist	O
at	O
DeepMind	B-Application
,	O
says	O
.	O
</s>
<s>
Greg	B-Application
Kohs	I-Application
,	O
the	O
director	O
of	O
the	O
film	O
,	O
said	O
“	O
The	O
complexity	O
of	O
the	O
game	O
of	O
Go	O
,	O
combined	O
with	O
the	O
technical	O
depth	O
of	O
an	O
emerging	O
technology	O
like	O
artificial	B-Application
intelligence	I-Application
seemed	O
like	O
it	O
might	O
create	O
an	O
insurmountable	O
barrier	O
for	O
a	O
film	O
like	O
this	O
.	O
</s>
<s>
The	O
fact	O
that	O
I	O
was	O
so	O
innocently	O
unaware	O
of	O
Go	O
and	O
AlphaGo	B-Application
actually	O
proved	O
to	O
be	O
beneficial	O
.	O
</s>
<s>
Greg	B-Application
Kohs	I-Application
also	O
said	O
that	O
“	O
Unlike	O
the	O
film	O
’s	O
human	O
characters	O
–	O
who	O
turn	O
their	O
curious	O
quest	O
for	O
knowledge	O
into	O
an	O
epic	O
spectacle	O
with	O
great	O
existential	O
implications	O
,	O
who	O
dare	O
to	O
risk	O
their	O
reputation	O
and	O
pride	O
to	O
contest	O
that	O
curiosity	O
–	O
AI	B-Application
might	O
not	O
yet	O
possess	O
the	O
ability	O
to	O
empathize	O
.	O
</s>
<s>
John	O
Defore	O
from	O
The	O
Hollywood	O
Reporter	O
,	O
a	O
famous	O
news	O
report	O
that	O
focuses	O
on	O
the	O
film	O
industry	O
mentioned	O
in	O
Rotten	O
Tomato	O
that	O
this	O
documentary	O
is	O
“	O
an	O
involving	O
sports-rivalry	O
doc	O
with	O
an	O
AI	B-Application
twist.	O
”	O
“	O
In	O
the	O
end	O
,	O
observers	O
wonder	O
if	O
AlphaGo	B-Application
’s	O
odd	O
variety	O
of	O
intuition	O
might	O
not	O
kill	O
Go	O
as	O
an	O
intellectual	O
pursuit	O
but	O
shift	O
its	O
course	O
,	O
forcing	O
the	O
game	O
’s	O
scholars	O
to	O
consider	O
it	O
from	O
new	O
angles	O
.	O
</s>
<s>
Director	O
Greg	B-Application
Kohs	I-Application
brings	O
out	O
all	O
of	O
the	O
drama	O
and	O
rush	O
of	O
these	O
games	O
to	O
show	O
the	O
selection	O
process	O
for	O
the	O
master	B-Application
of	O
Go	O
.	O
</s>
<s>
On	O
this	O
small	O
stage	O
,	O
the	O
human	O
Go	O
champion	O
and	O
the	O
AI	B-Application
challenger	O
engage	O
in	O
an	O
intense	O
duel	O
.	O
</s>
<s>
According	O
to	O
Rechtshaffen	O
Michael	O
of	O
the	O
Los	O
Angeles	O
Times	O
,	O
who	O
gave	O
the	O
movie	O
80	O
out	O
of	O
100	O
possible	O
points	O
on	O
Metascore	O
,	O
he	O
said	O
:	O
“	O
It	O
helps	O
matters	O
when	O
you	O
have	O
a	O
group	O
of	O
engaging	O
human	O
subjects	O
like	O
soft-spoken	O
Sedol	O
,	O
who	O
’s	O
as	O
intensively	O
contemplative	O
as	O
the	O
game	O
itself	O
,	O
contrasted	O
by	O
the	O
spirited	O
,	O
personable	O
Fan	O
Hui	O
,	O
the	O
Paris-based	O
European	O
champ	O
who	O
accepts	O
an	O
offer	O
to	O
serve	O
as	O
an	O
advisor	O
for	O
the	O
DeepMind	B-Application
team	O
after	O
suffering	O
a	O
demoralizing	O
AI	B-Application
trouncing	O
”	O
.	O
</s>
<s>
Facebook	B-Application
has	O
also	O
been	O
working	O
on	O
its	O
own	O
Go-playing	O
system	O
darkforest	B-Application
,	O
also	O
based	O
on	O
combining	O
machine	O
learning	O
and	O
Monte	B-Application
Carlo	I-Application
tree	I-Application
search	I-Application
.	O
</s>
<s>
Although	O
a	O
strong	O
player	O
against	O
other	O
computer	B-Application
Go	I-Application
programs	O
,	O
as	O
of	O
early	O
2016	O
,	O
it	O
had	O
not	O
yet	O
defeated	O
a	O
professional	O
human	O
player	O
.	O
</s>
<s>
Darkforest	B-Application
has	O
lost	O
to	O
CrazyStone	B-Application
and	O
Zen	B-Application
and	O
is	O
estimated	O
to	O
be	O
of	O
similar	O
strength	O
to	O
CrazyStone	B-Application
and	O
Zen	B-Application
.	O
</s>
<s>
DeepZenGo	B-Application
,	O
a	O
system	O
developed	O
with	O
support	O
from	O
video-sharing	O
website	O
Dwango	O
and	O
the	O
University	O
of	O
Tokyo	O
,	O
lost	O
2	O
–	O
1	O
in	O
November	O
2016	O
to	O
Go	O
master	B-Application
Cho	O
Chikun	O
,	O
who	O
holds	O
the	O
record	O
for	O
the	O
largest	O
number	O
of	O
Go	O
title	O
wins	O
in	O
Japan	O
.	O
</s>
<s>
A	O
2018	O
paper	O
in	O
Nature	O
cited	O
AlphaGo	B-Application
's	O
approach	O
as	O
the	O
basis	O
for	O
a	O
new	O
means	O
of	O
computing	O
potential	O
pharmaceutical	O
drug	O
molecules	O
.	O
</s>
<s>
AlphaGo	B-Application
Master	I-Application
(	O
white	O
)	O
v	O
.	O
Tang	O
Weixing	O
(	O
31	O
December	O
2016	O
)	O
,	O
AlphaGo	B-Application
won	O
by	O
resignation	O
.	O
</s>
<s>
The	O
documentary	O
film	O
AlphaGo	B-Application
raised	O
hopes	O
that	O
Lee	O
Sedol	O
and	O
Fan	O
Hui	O
would	O
have	O
benefitted	O
from	O
their	O
experience	O
of	O
playing	O
AlphaGo	B-Application
,	O
but	O
as	O
of	O
May	O
2018	O
their	O
ratings	O
were	O
little	O
changed	O
;	O
Lee	O
Sedol	O
was	O
ranked	O
11th	O
in	O
the	O
world	O
,	O
and	O
Fan	O
Hui	O
545th	O
.	O
</s>
<s>
On	O
19	O
November	O
2019	O
,	O
Lee	O
announced	O
his	O
retirement	O
from	O
professional	O
play	O
,	O
arguing	O
that	O
he	O
could	O
never	O
be	O
the	O
top	O
overall	O
player	O
of	O
Go	O
due	O
to	O
the	O
increasing	O
dominance	O
of	O
AI	B-Application
.	O
</s>
