we have a new superpower it’s directing
the evolution directing devolution is
something that elect has created new
amazing things far beyond the limit of
the theory when you usually want to
create something new you have to know
how to do it
for example behind the creation of a
skyscraper there are the laws of statics
behind the creation of a new chemical
product there are the laws of chemistry
and behind the creation of a new
computer application there are the
underlying procedures or algorithm
theory is essential nowaday to create
new things theory is a sort of
generalization of the phenomenon and
from the general laws we can find the
solution to particular problems with the
theory we can rationally explain each
component of our system why is there and
how it works but what if we don’t know
the theory well we are in the dark
forest of ignorance and this is like the
situation we are in front of a true or
false examination and we studied nothing
we are totally unprepared so our only
choice is to do some random guessing
true false false true false obviously
this is the opposite of a rational
approach but after all this is our only
choice in this case is enough to study a
little bit and apply but there are
real-world situations for which there is
a real lack of theory situation
involving very complex structures or
maybe
dealing with a million or even billion
or free parameters situations that are
mysterious or only partially explained
for example we would like to create a
very complex structure constrained by
demanding constraint or maybe we would
like to synthesize a new kind of protein
but we don’t know the corresponding DNA
sequence that can codify this protein or
maybe we would like to create a new form
of intelligence but we cannot train the
neural network and because we don’t know
the underlying procedures explicitly so
what can we do the idea is that before
seeing the magic power coming from the
our new superpower let’s see the huge
generative force of the evolution me you
you you you everyone here is here
because of the generative force of
devolution and it may be seeking trivial
but from a situation of total disorder
evolution creates order and again maybe
is trivial but evolution created amazing
things without a theory without a theory
for example sequoias are big structures
but has been created through an
evolutionary process and without a
theory
so the photosynthesis is a very complex
chemical reaction and as being created
without a theory and also forms of
intelligence are that are populating our
world for example here we can see some
ends are created through continuous try
& retry and also obviously the
intelligence of great mammals or the
human brain so what is the secret the
secret is time plus freedom
time because there are no time
constraints for devolution
all the time is available and freedom
freedom to make it right
freedom to do mistakes freedom to do it
right to change the rules to overwrite
the rules and so on trial and error are
commonly see as a key element in
personal growth as we have seen and is
also a key element in all
intrapreneurial cultures also the
scientific method brings with itself a
component of try & retry so now let’s
see how we can exploit the force of
evolution and the idea is that to create
something new we need a theory there are
situation with where there is no theory
and evolution can generate amazing thing
without ative this idea is why don’t we
exploit this and this superpower is
directing devolution
so let’s now imagine that we would like
to create a sort of time bubble inside
which the time go very very fast and
think this world from the outside you
can see the links happens very very fast
inside it so from our perspective is a
very fast world it is also isolated
because there are no connection between
the inside of the world with the outside
and very bad things could happen inside
without any direct consequences on our
world and the third there is a an
environment an environment that is an
environment that is a place it initially
with the some particular features for
example could contain a good part and a
non-state part and may be a basic set of
rules that can partially limit the
freedom of the agent and the agent free
agents are the second component of our
system
and they are free to evolve and that
could be of one type of multiple types
or maybe also challenging each other so
they are constantly pushing up they are
linked and what about the direction the
direction comes from letting the region
give a sense give a govern and so from
how we select the features at the
beginning agent and environment we
somehow direct how they will evolve so
let’s see some practical examples
first of all generative design
generative design is very interesting
because from some user-defined calls
sorry for example I would like to
sympathise SP chair I specified that the
weight the cost and so on and then a
computer implemented method generates
the chair and not only one million or
even billion of opportunities of designs
that all satisfy our initial goal our
initial constraints and a very similar
procedure could be applied for example
in another completely different area for
example to synthetize new kind of
proteins and this method is a directed
evolution and from some user-defined
goals to synthetize for example a
protein for which we don’t know how to
codify it because maybe is not present
in nature carbon silicon protein for
example okay the we define what we want
and through million or even billion of
mutagenesis we can reach a sort of
genealogical tree and select the our
final goal our final result and finally
there is a third application for example
in the
field of the artificial intelligence
let’s suppose that we would like to
teach a robot to walk we don’t know how
to explicitly codify the working
procedure so what we can do we can
create a sort of simulation environment
where more and more generation are
trying to work
and finally we select the generation
that is more performing in the task of
work and we can iterate several time
without a deal with evolutionary method
so we are Volta robots my company design
intelligence for small and MANET
vehicles at CES Las Vegas this year we
presented a self-driving Rover that can
find its way through the forest relying
only with one front-facing camera no
other sensor so there is a form of
intelligence that process the data from
the world and convert the pixels into a
variable action and this happens also on
very different types of terrain so the
robot recognize what is a train what is
the force tray and this is amazing is
possible only in the last a few years
and now that we have seen these
tremendous superpower of directing
devolution let’s speak about the
possible implications and consequences
this means first of all two things first
house is traditionally the enemy of a
rational approach in the theory
formalization but here house is a
developing factor second not necessarily
the theory come before the creation of
new things in this case most of the time
come after the creation of something
first week
and then we formalized a theory if
possible and we must expect a huge
growth of this kind of evolutionary
methods in the next years in the number
of application fields in the scale we
must expect bigger and bigger
environments and with a lot of variables
in the growth in complexity and also the
level of the assignment will grow we
will give to our evolutionary systems
micro walls higher and higher level
questions and they will feedback us the
consequences are mainly two the first
one is that we absolutely need a
unifying philosophy across the values
application area that are implementing
nowadays this method because we need a
unifying philosophy that keeps all the
advances together and secondly the main
is acquiring more and more central
position in the universe and this
central position is the position of the
director and this is a huge
responsibility for for us because
director comes from Latin the region I
give a direction give a sense govern and
we must ask as where are we going is
more important than giving an answer is
more important to ask us where what we
want to achieve where are we going as
you see before all the methods start
when there is a user-defined goal new
amazing things new amazing objects a
biological object like proteins or even
form of life or even a new kind of
structures design are going to populate
our world
that’s a new form of intelligence they
turn somehow Allen and for this reason
all this kind of objects are some
somehow like a dream they are not fully
consciously understood in their inner
workings but actually they work thank
you
[Applause]
the evolution directing devolution is
something that elect has created new
amazing things far beyond the limit of
the theory when you usually want to
create something new you have to know
how to do it
for example behind the creation of a
skyscraper there are the laws of statics
behind the creation of a new chemical
product there are the laws of chemistry
and behind the creation of a new
computer application there are the
underlying procedures or algorithm
theory is essential nowaday to create
new things theory is a sort of
generalization of the phenomenon and
from the general laws we can find the
solution to particular problems with the
theory we can rationally explain each
component of our system why is there and
how it works but what if we don’t know
the theory well we are in the dark
forest of ignorance and this is like the
situation we are in front of a true or
false examination and we studied nothing
we are totally unprepared so our only
choice is to do some random guessing
true false false true false obviously
this is the opposite of a rational
approach but after all this is our only
choice in this case is enough to study a
little bit and apply but there are
real-world situations for which there is
a real lack of theory situation
involving very complex structures or
maybe
dealing with a million or even billion
or free parameters situations that are
mysterious or only partially explained
for example we would like to create a
very complex structure constrained by
demanding constraint or maybe we would
like to synthesize a new kind of protein
but we don’t know the corresponding DNA
sequence that can codify this protein or
maybe we would like to create a new form
of intelligence but we cannot train the
neural network and because we don’t know
the underlying procedures explicitly so
what can we do the idea is that before
seeing the magic power coming from the
our new superpower let’s see the huge
generative force of the evolution me you
you you you everyone here is here
because of the generative force of
devolution and it may be seeking trivial
but from a situation of total disorder
evolution creates order and again maybe
is trivial but evolution created amazing
things without a theory without a theory
for example sequoias are big structures
but has been created through an
evolutionary process and without a
theory
so the photosynthesis is a very complex
chemical reaction and as being created
without a theory and also forms of
intelligence are that are populating our
world for example here we can see some
ends are created through continuous try
& retry and also obviously the
intelligence of great mammals or the
human brain so what is the secret the
secret is time plus freedom
time because there are no time
constraints for devolution
all the time is available and freedom
freedom to make it right
freedom to do mistakes freedom to do it
right to change the rules to overwrite
the rules and so on trial and error are
commonly see as a key element in
personal growth as we have seen and is
also a key element in all
intrapreneurial cultures also the
scientific method brings with itself a
component of try & retry so now let’s
see how we can exploit the force of
evolution and the idea is that to create
something new we need a theory there are
situation with where there is no theory
and evolution can generate amazing thing
without ative this idea is why don’t we
exploit this and this superpower is
directing devolution
so let’s now imagine that we would like
to create a sort of time bubble inside
which the time go very very fast and
think this world from the outside you
can see the links happens very very fast
inside it so from our perspective is a
very fast world it is also isolated
because there are no connection between
the inside of the world with the outside
and very bad things could happen inside
without any direct consequences on our
world and the third there is a an
environment an environment that is an
environment that is a place it initially
with the some particular features for
example could contain a good part and a
non-state part and may be a basic set of
rules that can partially limit the
freedom of the agent and the agent free
agents are the second component of our
system
and they are free to evolve and that
could be of one type of multiple types
or maybe also challenging each other so
they are constantly pushing up they are
linked and what about the direction the
direction comes from letting the region
give a sense give a govern and so from
how we select the features at the
beginning agent and environment we
somehow direct how they will evolve so
let’s see some practical examples
first of all generative design
generative design is very interesting
because from some user-defined calls
sorry for example I would like to
sympathise SP chair I specified that the
weight the cost and so on and then a
computer implemented method generates
the chair and not only one million or
even billion of opportunities of designs
that all satisfy our initial goal our
initial constraints and a very similar
procedure could be applied for example
in another completely different area for
example to synthetize new kind of
proteins and this method is a directed
evolution and from some user-defined
goals to synthetize for example a
protein for which we don’t know how to
codify it because maybe is not present
in nature carbon silicon protein for
example okay the we define what we want
and through million or even billion of
mutagenesis we can reach a sort of
genealogical tree and select the our
final goal our final result and finally
there is a third application for example
in the
field of the artificial intelligence
let’s suppose that we would like to
teach a robot to walk we don’t know how
to explicitly codify the working
procedure so what we can do we can
create a sort of simulation environment
where more and more generation are
trying to work
and finally we select the generation
that is more performing in the task of
work and we can iterate several time
without a deal with evolutionary method
so we are Volta robots my company design
intelligence for small and MANET
vehicles at CES Las Vegas this year we
presented a self-driving Rover that can
find its way through the forest relying
only with one front-facing camera no
other sensor so there is a form of
intelligence that process the data from
the world and convert the pixels into a
variable action and this happens also on
very different types of terrain so the
robot recognize what is a train what is
the force tray and this is amazing is
possible only in the last a few years
and now that we have seen these
tremendous superpower of directing
devolution let’s speak about the
possible implications and consequences
this means first of all two things first
house is traditionally the enemy of a
rational approach in the theory
formalization but here house is a
developing factor second not necessarily
the theory come before the creation of
new things in this case most of the time
come after the creation of something
first week
and then we formalized a theory if
possible and we must expect a huge
growth of this kind of evolutionary
methods in the next years in the number
of application fields in the scale we
must expect bigger and bigger
environments and with a lot of variables
in the growth in complexity and also the
level of the assignment will grow we
will give to our evolutionary systems
micro walls higher and higher level
questions and they will feedback us the
consequences are mainly two the first
one is that we absolutely need a
unifying philosophy across the values
application area that are implementing
nowadays this method because we need a
unifying philosophy that keeps all the
advances together and secondly the main
is acquiring more and more central
position in the universe and this
central position is the position of the
director and this is a huge
responsibility for for us because
director comes from Latin the region I
give a direction give a sense govern and
we must ask as where are we going is
more important than giving an answer is
more important to ask us where what we
want to achieve where are we going as
you see before all the methods start
when there is a user-defined goal new
amazing things new amazing objects a
biological object like proteins or even
form of life or even a new kind of
structures design are going to populate
our world
that’s a new form of intelligence they
turn somehow Allen and for this reason
all this kind of objects are some
somehow like a dream they are not fully
consciously understood in their inner
workings but actually they work thank
you
[Applause]
Please follow and like us: