The following is a conversation with Keoki Jackson.

He’s the CTO of Lockheed Martin,

a company that through its long history

has created some of the most incredible engineering marvels

human beings have ever built,

including planes that fly fast and undetected,

defense systems that intersect nuclear threats

that can take the lives of millions,

and systems that venture out into space,

the moon, Mars, and beyond.

And these days, more and more artificial intelligence

has an assistive role to play in these systems.

I’ve read several books in preparation for this conversation.

It is a difficult one,

because in part Lockheed Martin builds military systems

that operate in a complicated world

that often does not have easy solutions

in the gray area between good and evil.

I hope one day this world will rid itself of war

in all its forms.

But the path to achieving that in a world

that does have evil is not obvious.

What is obvious is good engineering

and artificial intelligence research

has a role to play on the side of good.

Lockheed Martin and the rest of our community

are hard at work at exactly this task.

We talk about these and other important topics

in this conversation.

Also, most certainly, both Keoki and I

have a passion for space,

us humans venturing out toward the stars.

We talk about this exciting future as well.

This is the Artificial Intelligence Podcast.

If you enjoy it, subscribe on YouTube,

give it five stars on iTunes, support it on Patreon,

or simply connect with me on Twitter at Lex Friedman,

spelled F R I D M A N.

And now, here’s my conversation with Keoki Jackson.

I read several books on Lockheed Martin recently.

My favorite in particular is by Ben Rich,

Carlos Concord’s personal memoir.

It gets a little edgy at times.

But from that, I was reminded that the engineers

at Lockheed Martin have created some of the most

incredible engineering marvels human beings have ever built

throughout the 20th century and the 21st.

Do you remember a particular project or system at Lockheed

or before that at the Space Shuttle Columbia

that you were just in awe at the fact that us humans

could create something like this?

You know, that’s a great question.

There’s a lot of things that I could draw on there.

When you look at the Skunk Works and Ben Rich’s book

in particular, of course, it starts off with basically

the start of the jet age and the P 80.

And I had the opportunity to sit next to one

of the Apollo astronauts, Charlie Duke, recently at dinner.

And I said, hey, what’s your favorite aircraft?

And he said, well, it was by far the F 104 Starfighter,

which was another aircraft that came out of Lockheed there.

It was the first Mach 2 jet fighter aircraft.

They called it the missile with a man in it.

And so those are the kinds of things I grew up hearing

stories about.

You know, of course, the SR 71 is incomparable

as kind of the epitome of speed, altitude,

and just the coolest looking aircraft ever.

So there’s a reconnaissance, that’s a plane.

That’s a, yeah, intelligence surveillance

and reconnaissance aircraft that was designed

to be able to outrun, basically go faster

than any air defense system.

But, you know, I’ll tell you, I’m a space junkie.

That’s why I came to MIT.

That’s really what took me ultimately to Lockheed Martin.

And I grew up, and so Lockheed Martin, for example,

has been essentially at the heart of every planetary mission,

like all the Mars missions we’ve had a part in.

And we’ve talked a lot about the 50th anniversary

of Apollo here in the last couple of weeks, right?

But remember, 1976, July 20th, again, National Space Days,

the landing of the Viking lander on the surface of Mars,

just a huge accomplishment.

And when I was a young engineer at Lockheed Martin,

I got to meet engineers who had designed, you know,

various pieces of that mission as well.

So that’s what I grew up on is these planetary missions,

the start of the space shuttle era,

and ultimately had the opportunity

to see Lockheed Martin’s part.

Lockheed Martin’s part, and we can maybe talk about

some of these here, but Lockheed Martin’s part

in all of these space journeys over the years.

Do you dream, and I apologize for getting philosophical

at times, or sentimental.

I do romanticize the notion of space exploration.

So do you dream of the day when us humans colonize

another planet like Mars, or a man, a woman,

a human being steps on Mars?

Absolutely, and that’s a personal dream of mine.

I haven’t given up yet on my own opportunity

to fly into space, but as, you know,

from the Lockheed Martin perspective,

this is something that we’re working towards every day.

And of course, you know, we’re building

the Orion spacecraft, which is the most sophisticated

human rated spacecraft ever built.

And it’s really designed for these deep space journeys,

you know, starting with the moon,

but ultimately going to Mars and being the platform,

you know, from a design perspective,

we call the Mars base camp to be able to take humans

to the surface, and then after a mission

of a couple of weeks, bring them back up safely.

And so that is something I want to see happen

during my time at Lockheed Martin.

So I’m pretty excited about that.

And I think, you know, once we prove that’s possible,

you know, colonization might be a little bit further out,

but it’s something that I’d hope to see.

So maybe you can give a little bit of an overview

of, so Lockheed Martin has partnered with a few years ago

with Boeing to work with the DOD and NASA

to build launch systems and rockets with the ULA.

What’s beyond that?

What’s Lockheed’s mission timeline,

long term dream in terms of space?

You mentioned the moon, I’ve heard you talk about asteroids.

As Mars, what’s the timeline?

What’s the engineering challenges

and what’s the dream long term?

Yeah, I think the dream long term is to have

a permanent presence in space beyond low earth orbit,

ultimately with a long term presence on the moon

and then to the planets, to Mars.

And… Sorry to interrupt on that.

So long term presence means…

Sustained and sustainable presence in an economy,

a space economy that really goes alongside that.

With human beings and being able to launch perhaps

from those, so like hop?

You know, there’s a lot of energy

that goes in those hops, right?

So I think the first step is being able to get there

and to be able to establish sustained bases, right?

And build from there.

And a lot of that means getting, as you know,

things like the cost of launch down

and you mentioned United Launch Alliance.

And so I don’t wanna speak for ULA,

but obviously they’re working really hard

to on their next generation of launch vehicles

to maintain that incredible mission success record

that ULA has, but ultimately continue

to drive down the cost and make the flexibility,

the speed and the access ever greater.

So what’s the missions that are in the horizon

that you could talk to?

Is there a hope to get to the moon?

Absolutely, absolutely.

I mean, I think you know this, or you may know this,

there’s a lot of ways to accomplish some of these goals.

And so that’s a lot of what’s in discussion today.

But ultimately the goal is to be able to establish a base

essentially in cislunar space that would allow

for ready transfer from orbit to the lunar surface

and back again.

And so that’s sort of that near term,

I say near term in the next decade or so vision,

starting off with a stated objective by this administration

to get back to the moon in the 2024, 2025 timeframe,

which is right around the corner here.

How big of an engineering challenge is that?

I think the big challenge is not so much to go,

but to stay, right?

And so we demonstrated in the 60s

that you could send somebody up,

do a couple of days of mission

and bring them home again successfully.

Now we’re talking about doing that,

I’d say more to, I don’t wanna say an industrial scale,

but a sustained scale, right?

So permanent habitation, regular reuse of vehicles,

the infrastructure to get things like fuel, air,

consumables, replacement parts,

all the things that you need to sustain

that kind of infrastructure.

So those are certainly engineering challenges,

there are budgetary challenges,

and those are all things

that we’re gonna have to work through.

The other thing, and I shouldn’t,

I don’t wanna minimize this,

I mean, I’m excited about human exploration,

but the reality is our technology

and where we’ve come over the last 40 years essentially

has changed what we can do with robotic exploration as well.

And to me, it’s incredibly thrilling,

and this seems like old news now,

but the fact that we have rovers driving around

the surface of Mars and sending back data

is just incredible.

The fact that we have satellites in orbit around Mars

that are collecting weather,

they’re looking at the terrain, they’re mapping,

all of these kinds of things on a continuous basis,

that’s incredible.

And the fact that you got the time lag, of course,

going to the planets,

but you can effectively have virtual human presence there

in a way that we have never been able to do before.

And now with the advent of even greater processing power,

better AI systems, better cognitive systems

and decision systems,

you put that together with the human piece

and we’ve really opened up the solar system

in a whole different way.

And I’ll give you an example, we’ve got OSIRIS REx,

which is a mission to the asteroid Bennu.

So the spacecraft is out there right now

on basically a year mapping activity

to map the entire surface of that asteroid in great detail.

You know, all autonomously piloted, right?

But the idea then that, and this is not too far away,

it’s gonna go in,

it’s got a sort of fancy vacuum cleaner with a bucket,

it’s gonna collect the sample off the asteroid

and then send it back here to Earth.

And so, you know, we have gone from sort of those

tentative steps in the 70s, you know,

early landings, video of the solar system

to now we’ve sent spacecraft to Pluto,

we have gone to comets and brought and intercepted comets,

we’ve brought stardust, you know, material back.

So that’s, we’ve gone far

and there’s incredible opportunity to go even farther.

So it seems quite crazy that this is even possible,

that can you talk a little bit about

what it means to orbit an asteroid

and with a bucket to try to pick up some soil samples?

Yeah, so part of it is just kind of the, you know,

these are the same kinds of techniques we use here on Earth

for high speed, high accuracy imagery,

stitching these scenes together and creating

essentially high accuracy world maps, right?

And so that’s what we’re doing, obviously,

on a much smaller scale with an asteroid.

But the other thing that’s really interesting,

you put together sort of that neat control

and, you know, data and imagery problem.

But the stories around how we designed the collection,

I mean, as essentially, you know,

this is the sort of the human ingenuity element, right?

That, you know, essentially had an engineer who had a,

one day he’s like, oh, starts messing around with parts,

vacuum cleaner, bucket, you know,

maybe we could do something like this.

And that was what led to what we call

the pogo stick collection, right?

Where basically a thing comes down,

it’s only there for seconds, does that collection,

grabs the, essentially blows the regolith material

into the collection hopper and off it goes.

It doesn’t really land almost.

It’s a very short landing.

Wow, that’s incredible.

So what is, in those, we talked a little bit more

about space, what’s the role of the human in all of this?

What are the challenges?

What are the opportunities for humans

as they pilot these vehicles in space?

And for humans that may step foot

on either the moon or Mars?

Yeah, it’s a great question because, you know,

I just have been extolling the virtues of robotic

and, you know, rovers, autonomous systems,

and those absolutely have a role.

I think the thing that we don’t know how to replace today

is the ability to adapt on the fly to new information.

And I believe that will come, but we’re not there yet.

There’s a ways to go.

And so, you know, you think back to Apollo 13

and the ingenuity of the folks on the ground

and on the spacecraft essentially cobbled together

a way to get the carbon dioxide scrubbers to work.

Those are the kinds of things that ultimately, you know,

and I’d say not just from dealing with anomalies,

but, you know, dealing with new information.

You see something and rather than waiting

20 minutes or half an hour, an hour

to try to get information back and forth,

but be able to essentially revector on the fly,

collect, you know, different samples,

take a different approach,

choose different areas to explore.

Those are the kinds of things that human presence enables

that is still a ways ahead of us on the AI side.

Yeah, there’s some interesting stuff

we’ll talk about on the teaming side here on Earth.

That’s pretty cool to explore.

And in space, let’s not leave the space piece out.

So what does teaming, what does AI and humans

working together in space look like?

Yeah, one of the things we’re working on

is a system called Maya, which is,

you think of it, so it’s an AI assistant.

In space. In space, exactly.

And you think of it as the Alexa in space, right?

But this goes hand in hand with a lot of other developments.

And so today’s world, everything is essentially model based,

model based systems engineering

to the actual digital tapestry that goes through the design,

the build, the manufacture, the testing,

and ultimately the sustainment of these system.

And so our vision is really that, you know,

when our astronauts are there around Mars,

you’re gonna have that entire digital library

of the spacecraft, of its operations, all the test data,

all the test data and flight data from previous missions

to be able to look and see if there are anomalous conditions

and tell the humans and potentially deal with that

before it becomes a bad situation

and help the astronauts work through those kinds of things.

And it’s not just, you know,

dealing with problems as they come up,

but also offering up opportunities

for additional exploration capability, for example.

So that’s the vision is that, you know,

these are gonna take the best of the human

to respond to changing circumstances

and rely on the best of AI capabilities

to monitor these, you know,

this almost infinite number of data points

and correlations of data points

that humans frankly aren’t that good at.

So how do you develop systems in space like this,

whether it’s Alexa in space or in general,

any kind of control systems,

any kind of intelligent systems

when you can’t really test stuff too much out in space?

It’s very expensive to test stuff.

So how do you develop such systems?

Yeah, that’s the beauty of this digital twin, if you will.

And of course, with Lockheed Martin,

we’ve over the past, you know, five plus decades

been refining our knowledge of the space environment,

of how materials behave, dynamics,

the controls, the radiation environments,

all of these kinds of things.

So we’re able to create very sophisticated models.

They’re not perfect, but they’re very good.

And so you can actually do a lot.

I spent part of my career, you know,

simulating communication spacecraft,

you know, missile warning spacecraft, GPS spacecraft

in all kinds of scenarios and all kinds of environments.

So this is really just taking that to the next level.

The interesting thing is that now

you’re bringing into that loop

a system depending on how it’s developed

that may be non deterministic,

it may be learning as it goes.

And in fact, we anticipate

that it will be learning as it goes.

And so that brings a whole new level of interest,

I guess, into how do you do verification and validation

of these non deterministic learning systems

in scenarios that may go out of the bounds

or the envelope that you have initially designed them to.

So had this system and its intelligence

has the same complexity,

some of the same complexity human does

and learns over time, it’s unpredictable

in certain kinds of ways in the,

so you still, you also have to model that

when you’re thinking about it.

So in your thoughts, it’s possible

to model the majority of situations,

the important aspects of situations here on earth

and in space enough to test stuff?

Yeah, this is really an active area of research

and we’re actually funding university research

in a variety of places, including MIT.

This is in the realm of trust and verification

and validation of I’d say autonomous systems in general

and then as a subset of that autonomous systems

that incorporate artificial intelligence capabilities.

And this is not an easy problem.

We’re working with startup companies,

we’ve got internal R&D, but our conviction is

that autonomy and more and more AI enabled autonomy

is gonna be in everything that Lockheed Martin develops

and fields and it’s gonna be retrofitting it.

Autonomy and AI are gonna be retrofit

into existing systems, they’re gonna be part

of the design for all of our future systems.

And so maybe I should take a step back

and say the way we define autonomy.

So we talk about autonomy essentially a system

that composes, selects and then executes decisions

with varying levels of human intervention.

And so you could think of no autonomy.

So this is essentially the human doing the task.

You can think of effectively partial autonomy

where the human is in the loop.

So making decisions in every case

about what the autonomous system can do.

Either in the cockpit or remotely.

Or remotely, exactly, but still in that control loop.

And then there’s what you’d call supervisory autonomy.

So the autonomous system is doing most of the work,

the human can intervene to stop it

or to change the direction.

And then ultimately full autonomy

where the human is off the loop altogether.

And for different types of missions

wanna have different levels of autonomy.

So now take that spectrum and this conviction

that autonomy and more and more AI

are in everything that we develop.

The kinds of things that Lockheed Martin does,

a lot of times are safety of life critical kinds of missions.

You think about aircraft, for example.

And so we require and our customers require

an extremely high level of confidence.

One, that we’re gonna protect life.

Two, that these systems will behave

in ways that their operators can understand.

And so this gets into that whole field.

Again, being able to verify and validate

that the systems have been and that they will operate

the way they’re designed and the way they’re expected.

And furthermore, that they will do that

in ways that can be explained and understood.

And that is an extremely difficult challenge.

Yeah, so here’s a difficult question.

I don’t mean to bring this up,

but I think it’s a good case study

that people are familiar with the Boeing 737 Max

commercial airplane has had two recent crashes

where their flight control software system failed

and it’s software.

So I don’t mean to speak about Boeing,

but broadly speaking, we have this

in the autonomous vehicle space too, semi autonomous.

We have millions of lines of code software making decisions.

There is a little bit of a clash of cultures

because software engineers don’t have the same culture

of safety often that people who build systems

like at Lockheed Martin do where it has to be

exceptionally safe, you have to test this on.

So how do we get this right when software

is making so many decisions?

Yeah, and there’s a lot of things that have to happen.

And by and large, I think it starts with the culture,

which is not necessarily something that A,

is taught in school or B is something that would come,

depending on what kind of software you’re developing,

it may not be relevant, right?

If you’re targeting ads or something like that.

So, and by and large, I’d say not just Lockheed Martin,

but certainly the aerospace industry as a whole

has developed a culture that does focus on safety,

safety of life, operational safety, mission success.

But as you note, these systems

have gotten incredibly complex.

And so they’re to the point where it’s almost impossible,

you know, state spaces become so huge

that it’s impossible to, or very difficult

to do a systematic verification across the entire set

of potential ways that an aircraft could be flown,

all the conditions that could happen,

all the potential failure scenarios.

Now, maybe that’s soluble one day,

maybe when we have our quantum computers

at our fingertips, we’ll be able to actually

simulate across an entire, you know,

almost infinite state space.

But today, you know, there’s a lot of work

to really try to bound the system,

to make sure that it behaves in predictable ways,

and then have this culture of continuous inquiry

and skepticism and questioning to say,

did we really consider the right realm of possibilities?

Have we done the right range of testing?

Do we really understand, you know, in this case,

you know, human and machine interactions,

the human decision process alongside the machine processes?

And so that’s that culture,

we call it the culture of mission success at Lockheed Martin

that really needs to be established.

And it’s not something, you know,

it’s something that people learn by living in it.

And it’s something that has to be promulgated, you know,

and it’s done, you know, from the highest levels

at a company of Lockheed Martin, like Lockheed Martin.

Yeah, and the same is being faced

at certain autonomous vehicle companies

where that culture is not there

because it started mostly by software engineers.

So that’s what they’re struggling with.

Is there lessons that you think we should learn

as an industry and a society from the Boeing 737 MAX crashes?

These crashes obviously are tremendous tragedies.

They’re tragedies for all of the people,

the crew, the families, the passengers,

the people on the ground involved.

And, you know, it’s also a huge business

and economic setback as well.

I mean, you know, we’ve seen that it’s impacting

essentially the trade balance of the US.

So these are important questions.

And these are the kinds that, you know,

we’ve seen similar kinds of questioning at times.

You know, you go back to the Challenger accident.

And it is, I think, always important to remind ourselves

that humans are fallible, that the systems we create,

as perfect as we strive to make them,

we can always make them better.

And so another element of that culture of mission success

is really that commitment to continuous improvement.

If there’s something that goes wrong,

a real commitment to root cause

and true root cause understanding,

to taking the corrective actions

and to making the future systems better.

And certainly we strive for, you know, no accidents.

And if you look at the record

of the commercial airline industry as a whole

and the commercial aircraft industry as a whole,

you know, there’s a very nice decaying exponential

to years now where we have

no commercial aircraft accidents at all, right?

Fatal accidents at all.

So that didn’t happen by accident.

It was through the regulatory agencies, FAA,

the airframe manufacturers really working on a system

to identify root causes and drive them out.

So maybe we can take a step back

and many people are familiar, but Lockheed Martin broadly,

what kind of categories of systems

are you involved in building?

You know, Lockheed Martin, we think of ourselves

as a company that solves hard mission problems.

And the output of that might be an airplane or a spacecraft

or a helicopter or a radar or something like that.

But ultimately we’re driven by these, you know,

what is our customer?

What is that mission that they need to achieve?

And so that’s what drove the SR71, right?

How do you get pictures of a place

where you’ve got sophisticated air defense systems

that are capable of handling any aircraft

that was out there at the time, right?

So that, you know, that’s what yielded an SR71.

Let’s build a nice flying camera.

Exactly.

And make sure it gets out and it gets back, right?

And that led ultimately to really the start

of the space program in the US as well.

So now take a step back to Lockheed Martin of today.

And we are, you know, on the order of 105 years old now

between Lockheed and Martin, the two big heritage companies.

Of course, we’re made up of a whole bunch

of other companies that came in as well.

General Dynamics, you know, kind of go down the list.

Today, you can think of us in this space

of solving mission problems.

So obviously on the aircraft side, tactical aircraft,

building the most advanced fighter aircraft

that the world has ever seen.

We’re up to now several hundred of those delivered,

building almost a hundred a year.

And of course, working on the things that come after that.

On the space side, we are engaged

in pretty much every venue of space utilization

and exploration you can imagine.

So I mentioned things like navigation and timing GPS,

communication satellites, missile warning satellites.

We’ve built commercial surveillance satellites.

We’ve built commercial communication satellites.

We do civil space.

So everything from human exploration

to the robotic exploration of the outer planets.

And keep going on the space front.

But a couple of other areas that I’d like to put out,

we’re heavily engaged in building

critical defensive systems.

And so a couple that I’ll mention, the Aegis Combat System.

This is basically the integrated air and missile defense

system for the US and allied fleets.

And so protects carrier strike groups, for example,

from incoming ballistic missile threats,

aircraft threats, cruise missile threats,

and kind of go down the list.

So the carriers, the fleet itself

is the thing that is being protected.

The carriers aren’t serving

as a protection for something else.

Well, that’s a little bit of a different application.

We’ve actually built the version called Aegis Ashore,

which is now deployed in a couple of places around the world.

So that same technology, I mean, basically can be used

to protect either an ocean going fleet

or a land based activity.

Another one, the THAAD program.

So THAAD, this is the Theater High Altitude Area Defense.

This is to protect relatively broad areas

against sophisticated ballistic missile threats.

And so now it’s deployed with a lot of US capabilities.

And now we have international customers

that are looking to buy that capability as well.

And so these are systems that defend,

not just defend militaries and military capabilities,

but defend population areas.

We saw maybe the first public use of these

back in the first Gulf War with the Patriot Systems.

And these are the kinds of things

that Lockheed Martin delivers.

And there’s a lot of stuff that goes into it.

A lot of stuff that goes with it.

So think about the radar systems and the sensing systems

that cue these, the command and control systems

that decide how you pair a weapon

against an incoming threat.

And then all the human and machine interfaces

to make sure that they can be operated successfully

in very strenuous environments.

Yeah, there’s some incredible engineering

that at every front, like you said.

So maybe if we just take a look at Lockheed history broadly,

maybe even looking at Skunk Works.

What are the biggest,

most impressive milestones of innovation?

So if you look at stealth, I would have called you crazy

if you said that’s possible at the time.

And supersonic and hypersonic.

So traveling at, first of all,

traveling at the speed of sound is pretty damn fast.

And supersonic and hypersonic,

three, four, five times the speed of sound.

That seems, I would also call you crazy

if you say you can do that.

So can you tell me how it’s possible

to do these kinds of things?

And is there other milestones and innovation

that’s going on that you can talk about?

Yeah.

Well, let me start on the Skunk Works saga.

And you kind of alluded to it in the beginning.

Skunk Works is as much an idea as a place.

And so it’s driven really by Kelly Johnson’s 14 principles.

And I’m not gonna list all 14 of them off,

but the idea, and this I’m sure will resonate

with any engineer who’s worked

on a highly motivated small team before.

The idea that if you can essentially have a small team

of very capable people who wanna work

on really hard problems, you can do almost anything.

Especially if you kind of shield them

from bureaucratic influences,

if you create very tight relationships with your customers

so that you have that team

and shared vision with the customer.

Those are the kinds of things that enable the Skunk Works

to do these incredible things.

And we listed off a number that you brought up stealth.

And I wish I could have seen Ben Rich with a ball bearing

rolling it across the desk to a general officer

and saying, would you like to have an aircraft

that has the radar cross section of this ball bearing?

Probably one of the least expensive

and most effective marketing campaigns

in the history of the industry.

So just for people that are not familiar,

the way you detect aircraft,

I’m sure there’s a lot of ways,

but radar for the longest time,

there’s a big blob that appears in the radar.

How do you make a plane disappear

so it looks as big as a ball bearing?

What’s involved in technology wise there?

What’s the broadly sort of the stuff you can speak about?

I’ll stick to what’s in Ben Rich’s book.

But obviously the geometry of how radar gets reflected

and the kinds of materials that either reflect or absorb

are kind of the couple of the critical elements there.

And it’s a cat and mouse game, right?

I mean, you know, radars get better,

stealth capabilities get better.

And so it’s a really a game

of continuous improvement and innovation there.

I’ll leave it at that.

Yeah, so the idea that something is essentially invisible

is quite fascinating.

But the other one is flying fast.

So speed of sound is 750, 60 miles an hour.

So supersonic is three, you know, Mach three,

something like that.

Yeah, we talk about the supersonic obviously,

and we kind of talk about that as that realm from Mach one

up through about Mach five and then hypersonic.

So, you know, high supersonic speeds would be past Mach five.

And you got to remember Lockheed Martin

and actually other companies have been involved

in hypersonic development since the late 60s.

You know, you think of everything from the X 15

to the space shuttle as examples of that.

I think the difference now is if you look around the world,

particularly the threat environment that we’re in today,

you’re starting to see, you know, publicly,

folks like the Russians and the Chinese

saying they have hypersonic weapons capability

that could threaten US and allied capabilities.

And also basically, you know, the claims are

these could get around defensive systems

that are out there today.

And so there’s a real sense of urgency.

You hear it from folks like the undersecretary of defense

for research and engineering, Dr. Mike Griffin,

and others in the department of defense that hypersonics

is something that’s really important to the nation

in terms of both parity, but also defensive capabilities.

And so that’s something that, you know, we’re pleased.

It’s something that Lockheed Martin’s, you know,

had a heritage in, we’ve invested R and D dollars

on our side for many years.

And we have a number of things going on

with various US government customers in that field today

that we’re very excited about.

So I would anticipate we’ll be hearing more about that

in the future from our customers.

And I’ve actually haven’t read much about this.

Probably you can’t talk about much of it at all,

but on the defensive side,

it’s a fascinating problem of perception

of trying to detect things that are really hard to see.

Can you comment on how hard that problem is

and how hard is it to stay ahead,

even if we go back a few decades,

stay ahead of the competition?

Well, maybe I’d, again, you gotta think of these

as ongoing capability development.

And so think back to the early days of missile defense.

So this would be in the 80s, the SDI program.

And in that timeframe, we proved and Lockheed Martin proved

that you could hit a bullet with a bullet, essentially,

and which is something that had never been done before

to take out an incoming ballistic missile.

And so that’s led to these incredible hit to kill

kinds of capabilities, PAC 3.

That’s the Patriot Advanced Capability Model 3

that Lockheed Martin builds,

the THAAD system that I talked about.

So now hypersonics, they’re different from ballistic systems.

And so we gotta take the next step in defensive capability.

I can, I’ll leave that there, but I can only imagine.

Now, let me just comment sort of as an engineer,

it’s sad to know that so much that Lockheed has done

in the past is classified or today,

and it’s shrouded in secrecy.

It has to be by the nature of the application.

So like what I do, so what we do here at MIT,

we would like to inspire young engineers, young scientists,

and yet in the Lockheed case,

some of that engineer has to stay quiet.

How do you think about that?

How does that make you feel?

Is there a future where more can be shown

or is it just the nature of this world

that it has to remain secret?

It’s a good question.

I think the public can see enough of,

and including students who may be in grade school,

high school, college today,

to understand the kinds of really hard problems

that we work on.

And I mean, look at the F35, right?

And obviously a lot of the detailed performance levels

are sensitive and controlled.

But we can talk about what an incredible aircraft this is,

supersonic, super cruise, kind of a fighter,

stealth capabilities.

It’s a flying information system in the sky

with data fusion, sensor fusion capabilities

that have never been seen before.

So these are the kinds of things that I believe,

these are the kinds of things that got me excited

when I was a student.

I think these still inspire students today.

And the other thing I’d say,

I mean, people are inspired by space.

People are inspired by aircraft.

Our employees are also inspired by that sense of mission.

And I’ll just give you an example.

I had the privilege to work

and lead our GPS programs for some time.

And that was a case where I actually worked on a program

that touches billions of people every day.

And so when I said, I worked on GPS,

everybody knew what I was talking about,

even though they didn’t maybe appreciate

the technical challenges that went into that.

But I’ll tell you, I got a briefing one time

from a major in the Air Force.

And he said, I go by callsign GIMP, GPS is my passion.

I love GPS.

And he was involved in the operational test of the system.

And he said, I was out in Iraq,

and I was on a helicopter, Blackhawk helicopter,

and I was bringing back a sergeant

and a handful of troops from a deployed location.

And he said, my job is GPS.

So I asked that sergeant,

and he’s beaten down and kind of half asleep.

And I said, what do you think about GPS?

And he brightened up, his eyes lit up,

and he said, well, GPS,

that brings me and my troops home every day.

I love GPS.

And that’s the kind of story where it’s like,

okay, I’m really making a difference here

in the kind of work.

So that mission piece is really important.

The last thing I’ll say is,

and this gets to some of these questions

around advanced technologies.

It’s not, they’re not just airplanes

and spacecraft anymore.

For people who are excited

about advanced software capabilities,

about AI, about bringing machine learning,

these are the things that we’re doing

to exponentially increase the mission capabilities

that go on those platforms.

And those are the kinds of things

that I think are more and more visible to the public.

Yeah, I think autonomy, especially in flight,

is super exciting.

Do you see a day, here we go, back into philosophy,

future when most fighter jets

will be highly autonomous to a degree

where a human doesn’t need to be in the cockpit

in almost all cases?

Well, I mean, that’s a world

that to a certain extent we’re in today.

Now these are remotely piloted aircraft, to be sure.

But we have hundreds of thousands of flight hours a year now

in remotely piloted aircraft.

And then if you take the F35,

there are huge layers, I guess,

in levels of autonomy built into that aircraft

so that the pilot is essentially more of a mission manager

rather than doing the data,

the second to second elements of flying the aircraft.

So in some ways it’s the easiest aircraft

in the world to fly.

And kind of a funny story on that.

So I don’t know if you know

how aircraft carrier landings work,

but basically there’s what’s called a tail hook

and it catches wires on the deck of the carrier.

And that’s what brings the aircraft

to a screeching halt, right?

And there’s typically three of these wires.

So if you miss the first, the second one,

you catch the next one, right?

And we got a little criticism.

I don’t know how true this story is,

but we got a little criticism.

The F35 is so perfect, it always gets the second wires.

We’re wearing out the wire because it always hits that one.

But that’s the kind of autonomy that just makes these,

essentially up levels what the human is doing

to more of that mission manager.

So much of that landing by the F35 is autonomous.

Well, it’s just, the control systems are such

that you really have dialed out the variability

that comes with all the environmental conditions.

You’re wearing it out.

So my point is to a certain extent,

that world is here today.

Do I think that we’re gonna see a day anytime soon

when there are no humans in the cockpit?

I don’t believe that.

But I do think we’re gonna see much more

human machine teaming, and we’re gonna see that much more

at the tactical edge.

And we did a demo, and you asked about

what the Skunk Works is doing these days.

And so this is something I can talk about,

but we did a demo with the Air Force Research Laboratory.

We called it Have Raider.

And so using an F16 as an autonomous wingman,

and we demonstrated all kinds of maneuvers

and various mission scenarios with the autonomous F16

being that so called loyal or trusted wingman.

And so those are the kinds of things that,

we’ve shown what is possible now.

Given that you’ve up leveled that pilot

to be a mission manager, now they can control

multiple other aircraft.

Think of them almost as extensions of your own aircraft

flying alongside with you.

So that’s another example of how this is really

coming to fruition.

And then I mentioned the landings,

but think about just the implications for humans

and flight safety, and this goes a little bit back

to the discussion we were having about

how do you continuously improve the level of safety

through automation while working through the complexities

that automation introduces.

So one of the challenges that you have

in high performance fighter aircraft is what’s called G lock.

So this is G induced loss of consciousness.

So you pull nine Gs, you’re wearing a pressure suit,

that’s not enough to keep the blood going to your brain,

you black out.

And of course that’s bad if you happen to be flying low,

near the deck and in an obstacle or terrain environment.

And so we developed a system in our aeronautics division

called Auto Gcast, so autonomous ground collision

avoidance system.

And we built that into the F16.

It’s actually saved seven aircraft, eight pilots already

in a relatively short time it’s been deployed.

It was so successful that the Air Force said,

hey, we need to have this in the F35 right away.

So we’ve actually done testing of that now on the F35.

And we’ve also integrated an autonomous

air collision avoidance system.

So think the air to air problem.

So now it’s the integrated collision avoidance system.

But these are the kinds of capabilities,

I wouldn’t call them AI.

I mean, they’re very sophisticated models

of the aircraft dynamics coupled with the terrain models

to be able to predict when essentially the pilot

is doing something that is gonna take the aircraft

or the pilot’s not doing something in this case.

But it just gives you an example of how autonomy

can be really a lifesaver in today’s world.

It’s like a autonomous automated emergency braking in cars.

But is there any exploration of perception of,

for example, detecting a G lock that the pilot is out?

So as opposed to perceiving the external environment

to infer that the pilot is out,

but actually perceiving the pilot directly.

Yeah, this is one of those cases

where you’d like to not take action

if you think the pilot’s there.

And it’s almost like systems that try to detect

if a driver’s falling asleep on the road, right?

With limited success.

So, I mean, this is what I call

the system of last resort, right?

Where if the aircraft has determined

that it’s going into the terrain, get it out of there.

And this is not something that we’re just doing

in the aircraft world.

And I wanted to highlight,

we have a technology we call Matrix,

but this is developed at Sikorsky Innovations.

The whole idea there is what we call optimal piloting.

So not optional piloting or unpiloted, but optimal piloting.

So an FAA certified system.

So you have a high degree of confidence.

It’s generally pretty deterministic.

So we know that it’ll do in different situations,

but effectively be able to fly a mission

with two pilots, one pilot, no pilots.

And you can think of it almost as like a dial

of the level of autonomy that you want,

but able, so it’s running in the background at all times

and able to pick up tasks,

whether it’s sort of autopilot kinds of tasks

or more sophisticated path planning kinds of activities

to be able to do things like, for example,

land on an oil rig in the North Sea

in bad weather, zero, zero conditions.

And you can imagine, of course,

there’s a lot of military utility to capability like that.

You could have an aircraft that you want to send out

for a crewed mission, but then at night,

if you want to use it to deliver supplies

in an unmanned mode, that could be done as well.

And so there’s clear advantages there.

But think about on the commercial side,

if you’re an aircraft taken,

you’re gonna fly out to this oil rig.

If you get out there and you can’t land,

then you gotta bring all those people back,

reschedule another flight,

pay the overtime for the crew that you just brought back

because they didn’t get where they were going,

pay for the overtime for the folks

that are out there in the oil rig.

This is real economic,

these are dollars and cents kinds of advantages

we’re bringing in the commercial world as well.

So here’s a difficult question from the AI space

that I would love it if you’re able to comment.

So a lot of this autonomy in AI you’ve mentioned just now

has this empowering effect.

One is the last resort, it keeps you safe.

The other is there’s a, with the teaming

and in general, assistive AI.

And I think there’s always a race.

So the world is full of, the world is complex.

It’s full of bad actors.

So there’s often a race to make sure

that we keep this country safe, right?

But with AI, there is a concern

that it’s a slightly different race.

Though there’s a lot of people in the AI space

that are concerned about the AI arms race.

That as opposed to the United States becoming,

having the best technology and therefore keeping us safe,

even we lose ability to keep control of it.

So this, the AI arms race getting away

from all of us humans.

So do you share this worry?

Do you share this concern

when we’re talking about military applications

that too much control and decision making capabilities

giving to software or AI?

Well, I don’t see it happening today.

And in fact, this is something from a policy perspective,

it’s obviously a very dynamic space,

but the Department of Defense has put quite a bit

of thought into that.

And maybe before talking about the policy,

I’ll just talk about some of the why.

And you alluded to it being a sort of a complicated

and a little bit scary world out there,

but there’s some big things happening today.

You hear a lot of talk now about a return

to great powers competition,

particularly around China and Russia with the US,

but there are some other big players out there as well.

And what we’ve seen is the deployment of some very,

I’d say concerning new weapon systems,

particularly with Russia and breaching

some of the IRBM,

Intermediate Range Ballistic Missile Treaties,

that’s been in the news a lot.

The building of islands, artificial islands

in the South China Sea by the Chinese

and then arming those islands.

The annexation of Crimea by Russia,

the invasion of Ukraine.

So there’s some pretty scary things.

And then you add on top of that,

the North Korean threat has certainly not gone away.

There’s a lot going on in the Middle East

with Iran in particular.

And we see this global terrorism threat has not abated.

So there are a lot of reasons to look for technology

to assist with those problems,

whether it’s AI or other technologies like hypersonics,

which we discussed.

So now let me give just a couple of hypotheticals.

So people react sort of in the second timeframe, right?

Photon hitting your eye to movement

is on the order of a few tenths of a second

kinds of processing time.

Roughly speaking,

computers are operating in the nanosecond timescale, right?

So just to bring home what that means,

a nanosecond to a second is like a second to 32 years.

So seconds on the battlefield,

in that sense, literally are lifetimes.

And so if you can bring an autonomous

or AI enabled capability

that will enable the human to shrink,

maybe you’ve heard the term the OODA loop.

So this whole idea that a typical battlefield decision

is characterized by observe.

So information comes in, orient.

How does that, what does that mean in the context?

Decide, what do I do about it?

And then act, take that action.

If you can use these capabilities to compress that OODA loop

to stay inside what your adversary is doing,

that’s an incredible powerful force on the battlefield.

That’s a really nice way to put it,

that the role of AI and computing in general

has a lot to benefit from just decreasing

from 32 years to one second,

as opposed to on the scale of seconds and minutes and hours

making decisions that humans are better at making.

And it actually goes the other way too.

So that’s on the short timescale.

So humans kind of work in the one second,

two seconds to eight hours.

After eight hours, you get tired,

you gotta go to the bathroom, whatever the case might be.

So there’s this whole range of other things.

Think about surveillance and guarding facilities.

Think about moving material, logistics, sustainment.

A lot of these, what they call dull, dirty

and dangerous things that you need

to have sustained activity,

but it’s sort of beyond the length of time

that a human can practically do as well.

So there’s this range of things that are critical

in military and defense applications

that AI and autonomy are particularly well suited to.

Now, the interesting question that you brought up is,

okay, how do you make sure that stays within human control?

So that was the context for now the policy.

And so there is a DOD directive called 3000.09

because that’s the way we name stuff in this world.

But I’d say it’s well worth reading.

It’s only a couple of pages long,

but it makes some key points.

And it’s really around making sure

that there’s human agency and control

over use of semi autonomous and autonomous weapons systems,

making sure that these systems are tested,

verified and evaluated in realistic,

real world type scenarios,

making sure that the people are actually trained

on how to use them,

making sure that the systems have human machine interfaces

that can show what state they’re in

and what kinds of decisions they’re making,

making sure that you’ve established doctrine

and tactics and techniques and procedures

for the use of these kinds of systems.

And so, and by the way, I mean, none of this is easy,

but I’m just trying to lay kind of the picture

of how the US has said,

this is the way we’re gonna treat AI and autonomous systems,

that it’s not a free for all.

And like there are rules of war and rules of engagement

with other kinds of systems,

think chemical weapons, biological weapons,

we need to think about the same sorts of implications.

And this is something that’s really important

for Lockheed Martin.

I mean, obviously we are a hundred percent complying

with our customer and the policies and regulations,

but I mean, AI is an incredible enabler,

say within the walls of Lockheed Martin

in terms of improving production efficiency,

doing helping engineers, doing generative design,

improving logistics, driving down energy costs.

I mean, there are so many applications,

but we’re also very interested in some of the elements

of ethical application within Lockheed Martin.

So we need to make sure that things like privacy

is taken care of, that we do everything we can

to drive out bias in AI enabled kinds of systems,

that we make sure that humans are involved in decisions,

that we’re not just delegating accountability to algorithms.

And so for us, it all comes back,

I talked about culture before,

and it comes back to sort of the Lockheed Martin culture

and our core values.

And so it’s pretty simple for us and do what’s right,

respect others, perform with excellence.

And now how do we tie that back to the ethical principles

will govern how AI is used within Lockheed Martin.

And we actually have a world, pretty,

so you might not know this,

but there are actually awards for ethics programs.

Lockheed Martin’s had a recognized ethics program

for many years.

And this is one of the things that our ethics team

is working with our engineering team on.

One of the miracles to me, perhaps a layman,

again, I was born in the Soviet Union.

So I have echoes, at least in my family history

of World War II and the Cold War.

Do you have a sense of why human civilization

has not destroyed itself through nuclear war,

so nuclear deterrence?

And thinking about the future,

does this technology have a role to play here?

And what is the long term future

of nuclear deterrence look like?

Yeah, this is one of those hard, hard questions.

And I should note that Lockheed Martin is both proud

and privileged to play a part in multiple legs

of our nuclear and strategic deterrent systems

like the Trident submarine launch ballistic missiles.

You talk about, is there still a possibility

that the human race could destroy itself?

I’d say that possibility is real.

But interestingly, in some sense,

I think the strategic deterrence have prevented

the kinds of incredibly destructive world wars

that we saw in the first half of the 20th century.

Now, things have gotten more complicated since that time

and since the Cold War.

It is more of a multipolar great powers world today.

Just to give you an example, back then,

there were, in the Cold War timeframe,

just a handful of nations

that had ballistic missile capability by last count.

And this is a few years old.

There’s over 70 nations today that have that.

Similar kinds of numbers

in terms of space based capabilities.

So the world has gotten more complex and more challenging

and the threats, I think, have proliferated

in ways that we didn’t expect.

The nation today is in the middle of a recapitalization

of our strategic deterrent.

I look at that as one of the most important things

that our nation can do.

What is involved in deterrence?

Is it being ready to attack

or is it the defensive systems that catch attacks?

A little bit of both.

And so it’s a complicated game theoretical kind of program.

But ultimately,

we are trying to prevent the use of any of these weapons.

And the theory behind prevention is that

even if an adversary uses a weapon against you,

you have the capability to essentially strike back

and do harm to them that’s unacceptable.

And so that will deter them from making use

of these weapons systems.

The deterrence calculus has changed, of course,

with more nations now having these kinds of weapons.

But I think from my perspective, it’s very important

to maintain a strategic deterrent.

You have to have systems that you know will work

when they’re required to work.

Now you know that they have to be adaptable

to a variety of different scenarios in today’s world.

And so that’s what this recapitalization of systems

that were built over previous decades,

making sure that they are appropriate, not just for today,

but for the decades to come.

So the other thing I’d really like to note

is strategic deterrence has a very different

character today.

We used to think of weapons of mass destruction

in terms of nuclear, chemical, biological.

And today we have a cyber threat.

We’ve seen examples of the use of cyber weaponry.

And if you think about the possibilities

of using cyber capabilities or an adversary attacking the US

to take out things like critical infrastructure,

electrical grids, water systems,

those are scenarios that are strategic in nature

to the survival of a nation as well.

So that is the kind of world that we live in today.

And part of my hope on this is one that we can also develop

technical or technological systems,

perhaps enabled by AI and autonomy,

that will allow us to contain and to fight back

against these kinds of new threats

that were not conceived when we first developed

our strategic deterrence.

Yeah, I know that Lockheed is involved in cyber,

so I saw that you mentioned that.

It’s an incredibly, nuclear almost seems easier than cyber

because there’s so many attack,

there’s so many ways that cyber can evolve

in such an uncertain future.

But talking about engineering with a mission,

I mean, in this case that you’re engineering systems

that basically save the world.

Well, like I said, we’re privileged to work

on some very challenging problems

for very critical customers here in the US

and with our allies abroad as well.

Lockheed builds both military and nonmilitary systems.

And perhaps the future of Lockheed

may be more in nonmilitary applications

if you talk about space and beyond.

I say that as a preface to a difficult question.

So President Eisenhower in 1961 in his farewell address

talked about the military industrial complex

and that it shouldn’t grow beyond what is needed.

So what are your thoughts on those words,

on the military industrial complex,

on the concern of growth of their developments

beyond what may be needed?

That where it may be needed is a critical phrase, of course.

And I think it is worth pointing out, as you noted,

that Lockheed Martin,

we are in a number of commercial businesses

from energy to space to commercial aircraft.

And so I wouldn’t neglect the importance

of those parts of our business as well.

I think the world is dynamic and there was a time,

and it doesn’t seem that long ago to me,

it was while I was a graduate student here at MIT

and we were talking about the peace dividend

at the end of the Cold War.

If you look at expenditure on military systems

as a fraction of GDP,

we’re far below peak levels of the past.

And to me, at least, it looks like a time

where you’re seeing global threats changing in a way

that would warrant relevant investments

in defensive capabilities.

The other thing I’d note,

for military and defensive systems,

it’s not quite a free market, right?

We don’t sell to people on the street.

And that warrants a very close partnership

between, I’d say, the customers and the people

that design, build, and maintain these systems

because of the very unique nature,

the very difficult requirements,

the very great importance on safety

and on operating the way they’re intended every time.

And so that does create,

and frankly, it’s one of Lockheed Martin’s great strengths

is that we have this expertise built up over many years

in partnership with our customers

to be able to design and build these systems

that meet these very unique mission needs.

Yeah, because building those systems is very costly,

there’s very little room for mistake.

I mean, it’s, yeah, just Ben Rich’s book and so on

just tells the story.

It’s nerve wracking just reading it.

If you’re an engineer, it reads like a thriller.

Okay, let me, let’s go back to space for a second.

I guess.

I’m always happy to go back to space.

So a few quick, maybe out there,

maybe fun questions, maybe a little provocative.

What are your thoughts on the efforts

of the new folks, SpaceX and Elon Musk?

What are your thoughts about what Elon is doing?

Do you see him as competition?

Do you enjoy competition?

What are your thoughts?

Yeah, first of all, certainly Elon,

I’d say SpaceX and some of his other ventures

are definitely a competitive force in the space industry.

And do we like competition?

Yeah, we do.

And we think we’re very strong competitors.

I think it’s, you know, competition is what the US

is founded on in a lot of ways

and always coming up with a better way.

And I think it’s really important

to continue to have fresh eyes coming in, new innovation.

I do think it’s important to have level playing fields.

And so you wanna make sure

that you’re not giving different requirements

to different players.

But, you know, I tell people, you know,

I spent a lot of time at places like MIT.

I’m gonna be at the MIT Beaverwork Summer Institute

over the weekend here.

And I tell people, this is the most exciting time

to be in the space business in my entire life.

And it is this explosion of new capabilities

that have been driven by things like the, you know,

the massive increase in computing power,

things like the massive increase in comms capabilities,

advanced and additive manufacturing

are really bringing down the barriers to entry in this field

and it’s driving just incredible innovation.

And it’s happening at startups,

but it’s also happening at Lockheed Martin.

You may not realize this, but Lockheed Martin,

working with Stanford actually built the first CubeSat

that was launched here out of the US

that was called QuakeSat.

And we did that with Stellar Solutions.

This was right around just after 2000, I guess.

And so we’ve been in that, you know,

from the very beginning.

And, you know, I talked about some of these,

like, you know, Maya and Orion,

but, you know, we’re in the middle of what we call smartsats

and software defined satellites

that can essentially restructure and remap their purpose,

their mission on orbit to give you almost, you know,

unlimited flexibility for these satellites

over their lifetimes.

So those are just a couple of examples,

but yeah, this is a great time to be in space.

Absolutely.

So Wright Brothers flew for the first time 116 years ago.

So now we have supersonic stealth planes

and all the technology we’ve talked about.

What innovations, obviously you can’t predict the future,

but do you see Lockheed in the next 100 years?

If you take that same leap,

how will the world of technology and engineering change?

I know it’s an impossible question,

but nobody could have predicted

that we could even fly 120 years ago.

So what do you think is the edge of possibility

that we’re going to be exploring in the next 100 years?

I don’t know that there is an edge.

I, you know, we’ve been around

for almost that entire time, right?

The Lockheed brothers and Glen L. Martin

starting their companies in the basement of a church

and an old service station.

We’re very different companies today

than we were back then, right?

And that’s because we’ve continuously reinvented ourselves

over all of those decades.

I think it’s fair to say, I know this for sure,

the world of the future, it’s gonna move faster,

it’s gonna be more connected,

it’s gonna be more autonomous,

and it’s gonna be more complex than it is today.

And so this is the world, you know,

as a CTO at Lockheed Martin that I think about,

what are the technologies that we have to invest in?

Whether it’s things like AI and autonomy,

you know, you can think about quantum computing,

which is an area that we’ve invested in

to try to stay ahead of these technological changes,

and frankly, some of the threats that are out there.

I believe that we’re gonna be out there in the solar system,

that we’re gonna be defending and defending well

against probably, you know, military threats

that nobody has even thought about today.

We are going to be, we’re gonna use these capabilities

to have far greater knowledge of our own planet,

the depths of the oceans, you know,

all the way to the upper reaches of the atmosphere

and everything out to the sun

and to the edge of the solar system.

So that’s what I look forward to,

and I’m excited, I mean, just looking ahead

in the next decade or so to the steps

that I see ahead of us in that time.

I don’t think there’s a better place to end,

Keoki, thank you so much.

Lex, it’s been a real pleasure,

and sorry it took so long to get up here,

but I’m glad we were able to make it happen.