several years ago my company did
aerospace research for NASA at Ames
Research Center in San Jose California
they have a very powerful air traffic
control control tower simulator that
they can use to evaluate improvements to
existing airports study new airports and
solve their traffic problems real
controllers to direct real pilots flying
this aircraft that you see in this video
the real pilots fly the simulator from
another room but they are at a computer
terminal and each one can control two to
four aircraft so there’s a lot of
traffic possibilities the interesting
thing about the facility is it can
simulate virtually any airport almost
anywhere in the world
that’s very cool stuff especially if
you’re researcher
the toughest problem in airports is not
landing aircraft and taking them off the
toughest problem is what do you do with
them on the ground there’s a lot of them
they all need to go get to where they’re
gonna go whether it’s takeoff or a gate
and so they you have to orchestrate that
so they don’t run into each other this
is really true in airports like Chicago
O’Hare Kennedy in New York and this is
Los Angeles that you’re looking at
they all have multiple runways they all
have many taxiways and those of you like
me who have waited on a taxiway to
either take off or get to a gate can
relate to the problem so orchestrating
this aircraft dance is a bit of a
challenge but a local researcher had
studied the problem and he felt he had a
fully automated solution and it worked
perfectly on his laboratory so they
brought it out to Ames put it on the
simulator and it worked perfectly then
they introduced the human element a
controller and a pilot and you get one
of you two head of me which is it
it didn’t take long for the simulation
that was chugging along to slow down and
ultimately it ground to a halt and
here’s the point the simulation the
automation worked exactly as it was
designed to do perfectly what the
designer didn’t do it didn’t take into
account the human decision process that
other real people go through and so the
captain the the guy flying the airplane
and the guy controlling the guy flying
the airplane who were used to being in
charge right didn’t trust the simulation
Trust is an important part of our daily
lives we trust our friends to be
reliable and supportive we trust our car
to start when we want it to and above
all we trust for all of our electronics
to be available to work every time we
want them to whether they’re charged or
not so how do we define trust what are
the factors well the dictionary
definition works pretty well for humans
it is it’s a firm belief and the
reliability truth ability of someone or
something not bad but the definition
that two of the most prominent human
factors researchers in trust and
automation have come up with is probably
more accurate for what we’re talking
about the attitude that an agent will
help an individual’s goals in the
situation characterized by uncertainty
and vulnerability the words to take away
our reliability and uncertainty trust
between humans is learned over time by
association and by reliable behavior but
it can be lost when those conditions are
no longer true there’s evidence that
most people believe that machines are
nearly perfect so that results in an
initial trust of
that’s pretty high however the moment
those machines don’t behave in a way
that humans recognize as being reliable
and appropriate that trust can be lost
the piece of technology is astonishing
but the ability that we have to adapt to
it lags way behind the pace of research
however is glacial awful so one of the
most challenging problems with
automation is stuff goes wrong this is
the cost of concordia 32 people lost
their lives
oh no on this this wreck and it happened
because the captain didn’t trust the
navigation of the of the boat took it
over took manual control and he steered
it into a rock okay that’s the serious
part it’s possible to trust automation
too much we love them but they take us
in 20 places sometimes it also makes for
good TV if any of you ever watch The
Office you may have seen in the episode
where Steve Carell is character Michael
trusted his GPS so literally he drove it
into a lake driverless cars can save
lives and uber Google the Ottawa
industry lots of people are involved but
uber the first car on your left had a
little problem it missed six I’m sorry
that’s six
red lights in a test in San Francisco
the Tesla in the middle there was a
famous Florida crash where a truck
pulled into the path of the car the car
could not recognize it react to it
quickly enough and the guy lost his life
he even didn’t take control of the
vehicle even though he was getting
visual warnings to do so
and the cute one on the right
originally by Google now it’s called way
mo and it is fully autonomous if any of
you know anything about autonomy this is
somewhere between level 4 and level 5
and that’s there’s the didactic part of
the talk that’s what that’s done with
that means that you don’t get to do
anything except get in turn it on when
you get there you turn it off and you
get out but it has been pulled over by a
cop
who was driving too slow pilotless
airplanes had the potential to have
safety benefits and economic benefits
this one happens to be from Airbus but
in a recent UBS survey only 17% of the
people responding to the survey said
they actually fly online and in truth
even if you would they’re still going to
have neat ground monitoring for the
longest period of time because somebody
needs to take over if something goes
wrong in the air trust and automation
can also evolve over time how many of
you like me counted your money
religiously the first few times you used
an ATM I’ll come on you know you did I
did I still do sometimes and but
ultimately we get to the point where you
realize they’re pretty reliably going to
work all the time so the common element
and all of these things is that few if
any designs take into account the human
and they particulate human need for
reduced uncertainty but an advance of
technology is going on no matter what so
finding a way to deal with this
uncertainty is going to be critical for
the future that requires more and better
research human focused research and a
lot of the human-computer human-machine
stuff is going on today in academia and
in in industry
Microsoft and Disney are two examples
but virtually none of it none of them
addresses trust trust and automation
research is however strong in the
Department of Defense in fact some is
going on in our backyard at
wright-patterson where the human
resource Laboratory has an office
specifically chartered to do that kind
of research the loyal wingman program is
a good example imagine if you will a
manned aircraft could be an f-35 but a
manned aircraft flying and joined in
formation by a
the unmanned aircraft fully automated
capable of breaking off and executing a
mission when this aircraft commands it
to do so yeah I thought so too
so clearly research in this whole area
is behind the power curve and this
lagging behind the pace of technology
development but wouldn’t our lives be so
much easier and more productive if we
could count on some human touch in the
automation we occur there may be some
hope this reconnaissance vessel is
called a sea hunter it is fully
autonomous it has human design elements
that were programmed into it from the
very beginning it is programmed to obey
all the laws of maritime operation and
can operate worldwide the difference is
that although it can do maritime
operation and it can determine right
away with between itself and other ships
in the area it does so with the
knowledge that it can maneuver out of
the way in a manner in which a human
would maneuver
thus conveying trust to the other ships
nearby so I’m gonna leave the last word
to Elon Musk where he said artificial
intelligence will be the best thing or
the worst thing ever for Humanity so
let’s get it right
aerospace research for NASA at Ames
Research Center in San Jose California
they have a very powerful air traffic
control control tower simulator that
they can use to evaluate improvements to
existing airports study new airports and
solve their traffic problems real
controllers to direct real pilots flying
this aircraft that you see in this video
the real pilots fly the simulator from
another room but they are at a computer
terminal and each one can control two to
four aircraft so there’s a lot of
traffic possibilities the interesting
thing about the facility is it can
simulate virtually any airport almost
anywhere in the world
that’s very cool stuff especially if
you’re researcher
the toughest problem in airports is not
landing aircraft and taking them off the
toughest problem is what do you do with
them on the ground there’s a lot of them
they all need to go get to where they’re
gonna go whether it’s takeoff or a gate
and so they you have to orchestrate that
so they don’t run into each other this
is really true in airports like Chicago
O’Hare Kennedy in New York and this is
Los Angeles that you’re looking at
they all have multiple runways they all
have many taxiways and those of you like
me who have waited on a taxiway to
either take off or get to a gate can
relate to the problem so orchestrating
this aircraft dance is a bit of a
challenge but a local researcher had
studied the problem and he felt he had a
fully automated solution and it worked
perfectly on his laboratory so they
brought it out to Ames put it on the
simulator and it worked perfectly then
they introduced the human element a
controller and a pilot and you get one
of you two head of me which is it
it didn’t take long for the simulation
that was chugging along to slow down and
ultimately it ground to a halt and
here’s the point the simulation the
automation worked exactly as it was
designed to do perfectly what the
designer didn’t do it didn’t take into
account the human decision process that
other real people go through and so the
captain the the guy flying the airplane
and the guy controlling the guy flying
the airplane who were used to being in
charge right didn’t trust the simulation
Trust is an important part of our daily
lives we trust our friends to be
reliable and supportive we trust our car
to start when we want it to and above
all we trust for all of our electronics
to be available to work every time we
want them to whether they’re charged or
not so how do we define trust what are
the factors well the dictionary
definition works pretty well for humans
it is it’s a firm belief and the
reliability truth ability of someone or
something not bad but the definition
that two of the most prominent human
factors researchers in trust and
automation have come up with is probably
more accurate for what we’re talking
about the attitude that an agent will
help an individual’s goals in the
situation characterized by uncertainty
and vulnerability the words to take away
our reliability and uncertainty trust
between humans is learned over time by
association and by reliable behavior but
it can be lost when those conditions are
no longer true there’s evidence that
most people believe that machines are
nearly perfect so that results in an
initial trust of
that’s pretty high however the moment
those machines don’t behave in a way
that humans recognize as being reliable
and appropriate that trust can be lost
the piece of technology is astonishing
but the ability that we have to adapt to
it lags way behind the pace of research
however is glacial awful so one of the
most challenging problems with
automation is stuff goes wrong this is
the cost of concordia 32 people lost
their lives
oh no on this this wreck and it happened
because the captain didn’t trust the
navigation of the of the boat took it
over took manual control and he steered
it into a rock okay that’s the serious
part it’s possible to trust automation
too much we love them but they take us
in 20 places sometimes it also makes for
good TV if any of you ever watch The
Office you may have seen in the episode
where Steve Carell is character Michael
trusted his GPS so literally he drove it
into a lake driverless cars can save
lives and uber Google the Ottawa
industry lots of people are involved but
uber the first car on your left had a
little problem it missed six I’m sorry
that’s six
red lights in a test in San Francisco
the Tesla in the middle there was a
famous Florida crash where a truck
pulled into the path of the car the car
could not recognize it react to it
quickly enough and the guy lost his life
he even didn’t take control of the
vehicle even though he was getting
visual warnings to do so
and the cute one on the right
originally by Google now it’s called way
mo and it is fully autonomous if any of
you know anything about autonomy this is
somewhere between level 4 and level 5
and that’s there’s the didactic part of
the talk that’s what that’s done with
that means that you don’t get to do
anything except get in turn it on when
you get there you turn it off and you
get out but it has been pulled over by a
cop
who was driving too slow pilotless
airplanes had the potential to have
safety benefits and economic benefits
this one happens to be from Airbus but
in a recent UBS survey only 17% of the
people responding to the survey said
they actually fly online and in truth
even if you would they’re still going to
have neat ground monitoring for the
longest period of time because somebody
needs to take over if something goes
wrong in the air trust and automation
can also evolve over time how many of
you like me counted your money
religiously the first few times you used
an ATM I’ll come on you know you did I
did I still do sometimes and but
ultimately we get to the point where you
realize they’re pretty reliably going to
work all the time so the common element
and all of these things is that few if
any designs take into account the human
and they particulate human need for
reduced uncertainty but an advance of
technology is going on no matter what so
finding a way to deal with this
uncertainty is going to be critical for
the future that requires more and better
research human focused research and a
lot of the human-computer human-machine
stuff is going on today in academia and
in in industry
Microsoft and Disney are two examples
but virtually none of it none of them
addresses trust trust and automation
research is however strong in the
Department of Defense in fact some is
going on in our backyard at
wright-patterson where the human
resource Laboratory has an office
specifically chartered to do that kind
of research the loyal wingman program is
a good example imagine if you will a
manned aircraft could be an f-35 but a
manned aircraft flying and joined in
formation by a
the unmanned aircraft fully automated
capable of breaking off and executing a
mission when this aircraft commands it
to do so yeah I thought so too
so clearly research in this whole area
is behind the power curve and this
lagging behind the pace of technology
development but wouldn’t our lives be so
much easier and more productive if we
could count on some human touch in the
automation we occur there may be some
hope this reconnaissance vessel is
called a sea hunter it is fully
autonomous it has human design elements
that were programmed into it from the
very beginning it is programmed to obey
all the laws of maritime operation and
can operate worldwide the difference is
that although it can do maritime
operation and it can determine right
away with between itself and other ships
in the area it does so with the
knowledge that it can maneuver out of
the way in a manner in which a human
would maneuver
thus conveying trust to the other ships
nearby so I’m gonna leave the last word
to Elon Musk where he said artificial
intelligence will be the best thing or
the worst thing ever for Humanity so
let’s get it right
Please follow and like us: