President Barack Obama on What AI Means for National Security

Some really positive outcomes,

but there are certainly some risks.

Certainly we've heard from some folks like Elon

and Mick Bustrom concerned about A.I.'s potential

to outpace our ability to understand it.

What about those concerns and how do we think

about that moving forward to protect

not only ourselves, but humanity at scale?

So let me start with what I think

is the more immediate concern,

that's a solvable problem, but we have to be mindful of it.

That is this category of specialized A.I.

If you've got a computer that can play Go,

it's a pretty complicated game with a lot of variations.

Developing an algorithm that simply says

maximize profits on the New York stock exchange,

is probably within sight.

If one person, or one organization, got there first,

they could bring down the stock market pretty quickly.

Or at least they could

raise questions about the integrity

of the financial markets.

An algorithm that said go figure out how

to penetrate the nuclear code in the country

and figure out how to launch some missiles.

If that's their only job, it's very narrow.

It doesn't require a super intelligence,

it just requires a really effective algorithm.

If it's self teaching, then you've got problems.

So part of, I think, my directive

to my national security team is

don't worry as much yet,

about machines taking over the world.

Do worry about

the capacity of either non-state actors

or hostile actors, to penetrate systems.

In that sense it's not

conceptually different, or different in a legal sense

than a lot of they cyber security work that we're doing.

It just means that we're gonna have to be better

because those who might deploy these systems

are going to get a lot better.

Now, I think as a precaution,

and all of us have spoken to folks

like Elon Musk who are concerned about

the super intelligent machine.

There's some prudence in thinking about benchmarks

that would indicate some general intelligence

developing on the horizon.

If we can see that coming

over the course of three decades, five decades,

you know whatever the late assessments are,

if ever, because there also are arguments

that this thing is a lot more complicated

than people make it out to be.

Then future generations, or our kids or our grandkids

are gonna be able to see it coming and figure it out.

But I do worry right now about specialized A.I.

I was on the West Coast and some kid

who looked like he was 25 shows me a laptop.

Or not a laptop, an iPad, and he says,

This is the future of radiology.

And he's got an algorithm that is teaching

enough sufficient pattern recognition

that over time, it's gonna be a better identifier

of disease than a radiologist would be.

If that's already happening today

on an iPad, you know invented by some kid at MIT,

then the vulnerability of a lot of our systems

is gonna be coming around pretty quick.

We're gonna have to have some preparation for that.

But Joey may have worse nightmares.

I generally agree.

The only caveat is, I would say, there are a few people

who believe generally A.I. will happen

at some fairly high percentage chance

in the next 10 years.

And these are people who are smart.

So I do think that keeping aware.

But the way I look at it is there's maybe a dozen

or two different breakthroughs that need to happen

for each of the pieces.

So you can kind of monitor it.

You don't know exactly when they're going to happen

because they're by definitions breakthroughs

and I think it's when you think

these breakthroughs will happen.

And you just have somebody close to the power cord.

(all laughing)

So right when you see it about to end, you gotta yank

that white piece out of the wall, man.

I'm completely with the President,

that short term, it's gonna be bad people

using A.I.'s for bad things

and they'll be an extension of us.

Then there's this other meta thing which happens

which is a group of people.

So if you look at all of the hate on the internet.

One person doesn't control that.

[President Obama] Right.

But it's a thing.

It points at things, it's definitely feeling

some political activity right now.

It's kind of got a life of it's own.

It's not even code, it's a culture.

And you see that also in the Middle East, right?

[President Obama] Which is why it's so hard to prevent.

Yeah, because it actually gets stronger

the more you attack it.

To me, what's curious and interesting

is going to be the relationship between an A.I.,

say a service that runs like that,

and then you throw in bitcoin, which is the ability

to move money around by a machine.

[Interviewer] Anonymously.

Anonymously, so to me, it will be this weird,

and again, this is where I think it could be embedded,

but if you gave this sort of mob, more tools.

'Cause they are actually fairly coordinated

in their own peculiar way.

On the good side is, you can imagine,

I was talking to some politicians like Michael Johnson

in Colorado, he's trying to figure out

how can we harness these things

to inform and engage citizens.

So to me, the problem is if you suppress it because of fear,

the bad guys will still use it.

What's important is to get people who want

to use it for good, communities and leaders,

and figure out how to get them to use it

so that's where we start to lean.

Yeah, this may not be a precise analogy.

Traditionally when we think about security

and protecting ourselves, we think in terms of

we need armor, or walls, from swords,

blunt instruments, et cetera.

Increasingly,

I find myself looking to medicine

and thinking about viruses, antibodies, right?

You know how do you create healthy systems,

that can ward off destructive elements?

In a distributed way.

In a distributed way and that requires more imagination

and we're not there yet.

It's part of the reason why cyber security

continues to be so hard.

Is because the threat is not a bunch of tanks

rolling at you, but a whole bunch of systems

that may be vulnerable to a worm getting in there.

It means that we've gotta think differently

about our security.

Make different investments

that may not be as sexy,

but actually may end up being

as important as anything.

Part of the reason I think about this

is because I also think that

what I spend a lot of time worrying about

are things like pandemic.

You can't build walls in order to prevent

the next airborne

lethal flu from landing on our shores.

Instead what we have to do

is be able to set up systems

to create public health systems

in all parts of the world,

quick triggers that tell us

when we see something emerging.

Make sure we've got quick protocols, systems,

that allow us to make vaccines a lot smarter.

So if you take that model,

a public health model,

when you think about how we can deal with

the problems of cyber security,

a lot of that may end up

being really helpful in thinking about

the A.I. threats.

And just one thing that I think is interesting,

is when we start to think about microbio, and microbes

everywhere, there's a lot of evidence to show

that introducing good bacteria to fight against

the bad bacteria is a strategy

and not to sterilize.

Well I still don't let Sonny and Bo lick me.

(all chuckling)

'Cause when I walk them on the South Lawn,

some of the things I see them do, ya know,

and chewing on I'm all like, hey man.

Stay away.

I think research has shown that actually opening windows

in hospitals instead of just

sterilizing the air may actually limit.

So we have to rethink what clean means.

It's similar whether you're talking about cyber security

or national security, I think the notion

that you can make straight borders

or that you can eliminate every possible pathogen

is difficult.

I think in that sense, in your position,

to be able to see medicine and cyber and A.I.,

I think that's an important thing.

Absolutely.

So there are distributed threats,

but is there also the risk that this creates

a new kind of arms race?

Look, I think there's no doubt that developing

international norms, rules,

protocols, verification mechanisms

around cyber security generally,

and A.I. in particular, is in it's infancy.

Part of the reason for that is,

as Joey identified, we've got a lot of non-state actors

who are the biggest players.

Part of the problem is that identifying

who is doing what is much more difficult.

If you're building a bunch of ICBM's,

we see 'em.

If somebody's sitting at a keyboard,

we don't.

So, we've begun this conversation.

A lot of the conversation right now

is not at the level of dealing with

real sophisticated A.I., but has more to do with

essentially states establishing norms

about how they use their cyber capabilities.

Part of what makes this an interesting problem

is that the line between offense and defense

is pretty blurred.

The truth of the matter is,

and part of the reason why, for example,

the debate here about cyber security.

Who are you more afraid of, big brother and the state?

Or the guy who's trying to empty out your bank account?

Part of the reason that's so difficult,

is that if we're going to police

this wild west,

whether it's the internet

or A.I. or any of these other areas,

then by definition, the government's

gotta have capabilities.

If it's got capabilities, then they're subject to abuse.

At a time when there's been a lot of mistrust built up,

about government, that makes it difficult.

When you have countries around the world

who see America as the preeminent cyber power,

now's the time for us to say,

we're willing to restrain ourselves,

if you are willing to restrain yourselves.

The challenge is the most sophisticated state actors,

Russia, China, Iran,

don't always embody the same norms or values that we do.

But we're gonna have to surface this

as an international issue in order for us to be effective.

'Cause effectively it's a borderless problem,

and ultimately, all states

are gonna have to worry about this.

It is very shortsighted if there's a state that thinks

that it can develop

super capacities in this area

without some 25 year old kid in a basement somewhere

figuring that out pretty quickly.