Please welcome Reid Hoffman.
-♪ -(cheering, applause)
-Welcome back.-Thank you.
-Welcome back to the show.-Excellent to be here.
Good to be here.
Good to be talkingabout artificial intelligence.
You people in Silicon Valley
are making the future.
People are terrifiedof the future.
When we talkabout artificial intelligence,
we have to break it downinto two groups, really, right?
And that isnarrow artificial intelligence,
which is, like, a chess computeror a self-driving car.
But then there's, like,artificial intelligence
like Arnold Schwarzenegger,right?
We want to talk about that.
What excites you mostabout artificial intelligence?
I think thatartificial intelligence
has the possibilityof curing disease,
it has the possibility of, uh,
essentially revolutionizingalmost every industry,
and it has the possibility
of essentially making, uh,productivity in every industry.
And so that can... as long aswe do it the right way
and involve peoplethe right way,
-Right. -I think that can bea really great future.
Give me an exampleof how it would revolutionize
-an industry. -Well, so, youalready mentioned, like, cars.
-Right. -So if you createautonomous vehicles,
-Uh-huh. -uh, essentially,it eliminates parking lots,
it eliminatestraffic congestion,
it essentially enablesmaybe a new set of suburbs
and construction,so that you can actually, uh,
go to different places,and you can either be working,
sleeping or partying whileyou're driving to and from work.
Why are you partyingwhile you're driving to work?
-(laughter) -What kind of workdo you work at, Reid?
-Well, I like my work.-I want to work at your office.
-I'm partyingon the way to work? -Yes.
Wow. Here's the thingI don't get, though.
You see, like, in that example,people always talk
about artificial intelligence,and, like, in that example,
I go, "But where are the cars?"
'Cause if we don't have parkinglots and no one owns the cars,
and, like, where are the cars?
Who owns the cars?
Well, there'smultiple configurations,
-Right. -but carscould just become a service.
So it's just like, for example,it's like Uber,
-Right. -but there's kind ofessentially cars in the cloud.
-Right. -Right?Various companies may own them.
And there's probably,way out of town,
there's probably parking lots,maintenance facilities,
-Uh-huh.-other kinds of things.
But they're not partof the, uh... of the...
You know, you can replace itwith parks, you can replace it
with, you know,office buildings,
-you know, residences.-Right.
It's not part of the inner city.
Okay, so that's the partof artificial intelligence
-I think we allsort of understand. -Yes.
Here's what everyonestruggles with.
The artificial intelligencethat we are warned about.
So, you're one of the peoplewho has come out and said,
"Hey, we've got to be cautious
"when it comesto artificial intelligence,
because it could kill us all."
-I'm paraphrasing you,obviously. -(laughing)
-But that's what I heard. -Idon't think I said exactly that.
-I heard it as "kill us all."-Yes.
Will artificial intelligencekill us all?
-Uh, very unlikely, I think.-Uh-huh.
-I think that the...-So possibly.
-Yeah. Well... Okay.-Okay. Okay.
Now I understandhow you heard this.
By the way, asteroids mightkill us all, too,
-just so you know.-Yes. Yes.
-And nuclear weapons mightkill us all. -Yes.
And I'm not making asteroidsfor a reason.
-But there are nuclearweapons.-Right, right, right.
No, no, no. But, but, but why,
why should we be wary ofartificial intelligence?
So the actual thing thateveryone worries about
-is the Terminator movie,-Right.
and Robocalypse, 'cause it's a nice story.
I think the likelihood ofthat story is pretty low.
Right? I mean, again, maybe anasteroid hitting the earth
-is more likely than that story.-Okay.
However, things like,for example, cyber security,
where artificial intelligencemight actually, in fact,
cause the, kind of,
attack and own the net.
And then cause problemswith the power grid
or other kinds of things.
Those kinds of thingsmake me more worried.
And it makes me more worriedif it's possibly an arms race
So those are, I think, the moredirect and immediate worries
in terms of where artificialintelligence applies.
So a Terminator that only wantsto shutdown my electricity?
Or, for example, disablesall the electricity,
but if you go withoutelectricity for a month,
-that's a problem.-Right.
-That is a huge problem.-Yes.
If you, if you're lookingat artificial intelligence
from, you know, at the--
from the height that you are,10,000 feet,
you're looking at it as a,as a global idea.
It is terrifying as a concept,
because one of the thingsthat gets to me is,
what happens to humans--not being killed off,
but rather in terms ofredefining our purpose.
So if robots are doing the jobs,
we've seen what that's doneto America already.
We see what it's doingall over the world.
Robots then plan the work,
robots then teach in schools,
robots then work ongenetic sequencing,
the robots are nowthe doctors as well.
What do humans do?What is our purpose?
Well, so I think, the problem--
I'll put it this way:
That's a possible universe,
-but there are many otherpossible universes. -Okay.
So for example, it's possiblethat I teach
-along with robots that help medo the teaching. -Okay.
Right? So it's not that therobots do all the teaching,
but it enables people to bemuch more productive.
-Interesting. -And so therefore,and there's a lot of use
in, like, uh, education,
in care for other people,
medical care, elderly care, etc.
You know, kind of, other kindsof person-to-person industry,
you could make thatmuch more intensive,
where it's stillpeople-to-people,
-being enabled by robots.-Right.
Now, that's just to illustratethat there's a possibility
that you can actually reallykind of tell a story about,
see that it's possible.
We also might haveinteresting entrepreneurship,
that also creates,like, for example,
much more, like, in personreal live entertainment,
where you come backto actually--
handmade goods, actually,get a lot of,
kind of, personal credit,status credit.
Like, you know,like people like clothing,
-and they like, and they like,you know, -Yes, yes, yes.
artifacts and bagsand everything else.
And that could actually still bepart of that future.
-It isn't necessarily that we'reall kind of, -I understand.
-sitting on couches, you know,watching television, -Right.
-and the robots are doingeverything else. -Got it.
Right? So there's a lotof futures. That's the...
Well, but-butyou cannot deny, though...
with the growthof artificial intelligence,
-there is threat to a jobas we know it. -Yes.
It happenswith every revolution.
It happens with every evolutionor every idea that changes.
So, with the telephoneand the exchanges--
you know, with the telephonebeing invented the way it was,
all of a suddenthe exchange broke apart.
The mobile phone changed that.We no longer needed
to ask somebodyto connect the call.
Now you could justget the call connected
-with the computer.-Yes.
-And then peoplelost their jobs. -Yes.
Now, more jobs were createdover time, but there's always
-that gap where there areno jobs. -Yes, 100%.
In Silicon Valley,are you considering this?
Are you thinkingabout all of the people
that would lose jobsin the interim--
the people who wouldl-lose a job as a driver,
the person who would losehis job as a truck driver,
the person who would losehis job as teacher's assistant
because nowthe robot is doing it?
Uh... not everyone'sthinking about it.
-But a bunch of peopleare thinking about it. -Right.
And they're thinking abouthow do we build A.I.s
-that also help people. -Right.
Right? Whether it's helping youre-skill for new jobs,
-whether it's helping youdo the job better. -Uh-huh.
Right? So it's both threads.There are people
who are thinking about,"Okay, how do I just make it
-so the robot cando the whole job?" -Okay.
All manufacturing. And thenthere's also people who say,
"How do I make the robotactually, in fact,
able to help peopledo the job much better?"
Okay. So the robotteaches me to do the job,
and another robotcan do the job.
And... W... Like, it...
Why would the robotnot want to make me not exist?
Like, why would the robotneed me now is what I'm saying.
Yeah. Well, uh...
the robot might need youfor a number of reasons.
So, one thing is, uh,there's a variety...
We don't knowwhat the current curve
in artificial intelligence--where it will, uh,
-kind of curve off.Is it person-to-person, -Uh-huh.
-the emotional connection?-Right.
-Is it a certain amountof creativity? -Uh-huh.
Um, is it the fact that the-therobot plus the person together
-actually, in fact,perform a lot better? -Right.
Right? That's also possible.Uh, there's a lot
of different options here.And, so, uh...
And then you get to the wholequestion of the "robot wanting."
Right? So... And this... youalluded this at the beginning,
which is, well, when do robotsbecome fully entities,
And what... Then what languageof want describes them?
And it's easyto tell human stories
where you project human wantsand not wants.
Like, "I can do this job.I don't need you, lousy human."
-Unclear that the robotwill react that way. -Right.
Like, how do we train it,uh, how do we, essentially,
bring it into it?And the way that I talk
to a lot of my colleaguesabout this--
and this maybe a little too abstract--
but it's how do wecreate symbiosis?
How do we make it so thatwe're both better together,
and that's actuallythe-the design target
for what we should be doing.
So, lookinginto the near future,
what is the onemost exciting thing
that we shouldall be looking forward to,
when it comes to artificialintelligence?
Uh, I think the most excitingthing is probably, uh,
uh, a much faster pathto curing diseases.
Because, actually, in fact,using A.I. to figure out
what's going on with thediseases, uh, which remedies
might work and how you mightfix it might actually, in fact,
uh, greatly, uh, expand humanlife across the planet.
Wow. It's exciting andfrightening at the same time.
Thank you so much for beingon the show. I appreciate it.
Be sure to listen to Reid'spodcast, Masters of Scale.
Reid Hoffman, everybody.