Don't fear superintelligent AI | Grady Booch

Prevodilac: Milenka Okuka Lektor: Mile Živković

When I was a kid, I was the quintessential nerd.

Kad sam bio dete, bio sam štreber do koske.

I think some of you were, too.

Mislim da su i neki od vas to bili.

(Laughter)

(Smeh)

And you, sir, who laughed the loudest, you probably still are.

A vi, gospodine, koji se najglasnije smejete, verovatno ste to i dalje.

(Laughter)

(Smeh)

I grew up in a small town in the dusty plains of north Texas,

Odrastao sam u gradiću u prašnjavim ravnicama severnog Teksasa

the son of a sheriff who was the son of a pastor.

kao sin šerifa koji je sin pastora.

Getting into trouble was not an option.

Upadanje u nevolju nije dolazilo u obzir.

And so I started reading calculus books for fun.

Te sam iz zabave počeo da čitam knjige o proračunu.

(Laughter)

(Smeh)

You did, too.

I vi ste.

That led me to building a laser and a computer and model rockets,

Zbog toga sam sagradio laser, kompjuter i modele raketa,

and that led me to making rocket fuel in my bedroom.

a to me je navelo da napravim raketno gorivo u spavaćoj sobi.

Now, in scientific terms,

Sad, u naučnom smislu,

we call this a very bad idea.

ovo nazivamo veoma lošom idejom.

(Laughter)

(Smeh)

Around that same time,

Otprilike u to vreme,

Stanley Kubrick's "2001: A Space Odyssey" came to the theaters,

u bioskope je stigla "Odiseja u svemiru: 2001" Stenlija Kjubrika,

and my life was forever changed.

i moj život je izmenjen zauvek.

I loved everything about that movie,

Sve mi se sviđalo kod tog filma,

especially the HAL 9000.

naročito HAL 9000.

Now, HAL was a sentient computer

Sad, HAL je bio svesni kompjuter

designed to guide the Discovery spacecraft

osmišljen da vodi kosmički brod Diskaveri

from the Earth to Jupiter.

od Zemlje do Jupitera.

HAL was also a flawed character,

HAL je takođe bio faličan lik

for in the end he chose to value the mission over human life.

jer je na kraju odabrao vrednost misije umesto ljudskog života.

Now, HAL was a fictional character,

Sad, HAL je bio fiktivan lik,

but nonetheless he speaks to our fears,

ali bez obzira na to, obraća se našim strahovima,

our fears of being subjugated

našim strahovima da ćemo pasti pod uticaj

by some unfeeling, artificial intelligence

nekakve bezosećajne, veštačke inteligencije

who is indifferent to our humanity.

koja je ravnodušna prema našoj ljudskosti.

I believe that such fears are unfounded.

Verujem da su takvi strahovi neosnovani.

Indeed, we stand at a remarkable time

Uistinu se nalazimo u izvanrednom vremenu

in human history,

u ljudskoj istoriji,

where, driven by refusal to accept the limits of our bodies and our minds,

u kom, pokrenuti odbijanjem da prihvatimo granice naših tela i naših umova,

we are building machines

gradimo mašine

of exquisite, beautiful complexity and grace

izuzetne, prelepe složenosti i stila

that will extend the human experience

koje će da prošire ljudsko iskustvo

in ways beyond our imagining.

na nezamislive načine.

After a career that led me from the Air Force Academy

Nakon karijere koja seže od akademije ratnog vazduhoplovstva

to Space Command to now,

preko svemirskog zapovednika do sada,

I became a systems engineer,

postao sam inženjer za sisteme,

and recently I was drawn into an engineering problem

a nedavno me je privukao inženjerski problem

associated with NASA's mission to Mars.

u vezi sa NASA-inom misijom na Marsu.

Now, in space flights to the Moon,

Sad, tokom svemirskih letova na Mesec

we can rely upon mission control in Houston

možemo da se oslonimo na kontrolu misije u Hjustonu

to watch over all aspects of a flight.

da će da prati sve aspekte leta.

However, Mars is 200 times further away,

Međutim, Mars je 200 puta dalje,

and as a result it takes on average 13 minutes

i zbog toga je potrebno u proseku 13 minuta

for a signal to travel from the Earth to Mars.

za signal da putuje od Zemlje do Marsa.

If there's trouble, there's not enough time.

Ako imamo problem, nema dovoljno vremena.

And so a reasonable engineering solution

Pa razumno inženjeresko rešenje

calls for us to put mission control

iziskuje od nas da smestimo kontrolu misije

inside the walls of the Orion spacecraft.

unutar zidova kosmičkog broda Orion.

Another fascinating idea in the mission profile

Još jedna fantastična ideja u profilu misije

places humanoid robots on the surface of Mars

smešta humanoidne robote na površinu Marsa

before the humans themselves arrive,

pre nego što ljudi lično stignu,

first to build facilities

prvo da bi gradili objekte

and later to serve as collaborative members of the science team.

i kasnije da služe kao saradnici u naučnoj ekipi.

Now, as I looked at this from an engineering perspective,

Sad, dok sam ovo posmatrao sa inženjerske tačke gledišta,

it became very clear to me that what I needed to architect

bilo mi je veoma jasno da je potrebno da projektujem

was a smart, collaborative,

pametnu, sarađivačku,

socially intelligent artificial intelligence.

društveno inteligentnu veštačku inteligenciju.

In other words, I needed to build something very much like a HAL

Drugim rečima, trebalo je da sagradim nešto prilično slično HAL-u,

but without the homicidal tendencies.

ali bez ubilačkih tendencija.

(Laughter)

(Smeh)

Let's pause for a moment.

Zastanimo na trenutak.

Is it really possible to build an artificial intelligence like that?

Da li je zaista moguće sagraditi takvu veštačku inteligenciju?

Actually, it is.

Zapravo jeste.

In many ways,

Na razne načine,

this is a hard engineering problem

ovo je težak inženjerski problem

with elements of AI,

sa elementima VI,

not some wet hair ball of an AI problem that needs to be engineered.

a ne nekakav neprijatan problem VI koji je potrebno projektovati.

To paraphrase Alan Turing,

Da parafraziram Alana Tjuringa:

I'm not interested in building a sentient machine.

ne zanima me izgradnja svesne mašine.

I'm not building a HAL.

Ne pravim HAL-a.

All I'm after is a simple brain,

Ja tragam za prostim mozgom,

something that offers the illusion of intelligence.

nečim što pruža iluziju inteligencije.

The art and the science of computing have come a long way

Umetnost nauke i računarstva su prešle dug put

since HAL was onscreen,

od vremena prikazivanja HAL-a,

and I'd imagine if his inventor Dr. Chandra were here today,

i pretpostavljam da je njegov tvorac dr Čandra ovde danas,

he'd have a whole lot of questions for us.

imao bi gomilu pitanja za nas.

Is it really possible for us

Da li je zaista moguće da mi

to take a system of millions upon millions of devices,

uzmemo sistem od miliona i miliona uređaja,

to read in their data streams,

koji učitava dotok podataka,

to predict their failures and act in advance?

i da predvidimo njihove kvarove i delamo unapred?

Yes.

Da.

Can we build systems that converse with humans in natural language?

Možemo li sagraditi sisteme koji pričaju s ljudima na prirodnom jeziku?

Yes.

Da.

Can we build systems that recognize objects, identify emotions,

Možemo li sagraditi sisteme koji prepoznaju predmete, osećanja,

emote themselves, play games and even read lips?

ispoljavaju emocije, igraju igrice, čak i čitaju sa usana?

Yes.

Da.

Can we build a system that sets goals,

Možemo li sagraditi sistem koji postavlja ciljeve,

that carries out plans against those goals and learns along the way?

koji pravi planove za te ciljeve i usput uči?

Yes.

Da.

Can we build systems that have a theory of mind?

Možemo li sagraditi sisteme koji sadrže teoriju uma?

This we are learning to do.

Ovo učimo da radimo.

Can we build systems that have an ethical and moral foundation?

Možemo li sagraditi sisteme koji imaju etičke i moralne osnove?

This we must learn how to do.

Ovo moramo da naučimo da radimo.

So let's accept for a moment

Zato, prihvatimo za trenutak

that it's possible to build such an artificial intelligence

da je moguće sagraditi takvu veštačku inteligenciju

for this kind of mission and others.

za ovakvu i slične misije.

The next question you must ask yourself is,

Sledeće pitanje koje morate postaviti sebi je:

should we fear it?

da li treba da je se plašimo?

Now, every new technology

Sad, svaka nova tehnologija

brings with it some measure of trepidation.

sa sobom nosi izvesnu dozu strepnje.

When we first saw cars,

Kada smo prvi put videli automobile,

people lamented that we would see the destruction of the family.

ljudi su se žalili da ćemo biti svedoci uništenja porodice.

When we first saw telephones come in,

Kada smo prvi put videli telefon,

people were worried it would destroy all civil conversation.

ljudi su bili zabrinuti da će da uništi učtiv razgovor.

At a point in time we saw the written word become pervasive,

U jednom trenutku smo videli kako pisana reč počinje da preovlađuje,

people thought we would lose our ability to memorize.

ljudi su smatrali da ćemo izgubiti sposobnost pamćenja.

These things are all true to a degree,

Sve ovo je donekle tačno,

but it's also the case that these technologies

ali je takođe tačno da nam je ova tehnologija

brought to us things that extended the human experience

donela stvari koje su proširile ljudsko iskustvo

in some profound ways.

na izvesne suštinske načine.

So let's take this a little further.

Stoga, pođimo malo dalje s ovim.

I do not fear the creation of an AI like this,

Ne strahujem od stvaranja VI na ovaj način

because it will eventually embody some of our values.

jer će vremenom otelotvoriti neke od naših vrednosti.

Consider this: building a cognitive system is fundamentally different

Razmotrite sledeće: izgradnja kognitivnog sistema je temeljno različita

than building a traditional software-intensive system of the past.

od izgradnje tradicionalnog softverskog sistema iz prošlosti.

We don't program them. We teach them.

Ne programiramo ih. Podučavamo ih.

In order to teach a system how to recognize flowers,

Kako bih naučio sistem da raspoznaje cveće,

I show it thousands of flowers of the kinds I like.

pokazujem mu na hiljade cvetova koji se meni sviđaju.

In order to teach a system how to play a game --

Kako bih naučio sistem da igra igru -

Well, I would. You would, too.

Pa, bih. I vi biste.

I like flowers. Come on.

Volim cveće. Pa šta.

To teach a system how to play a game like Go,

Kako bih naučio sistem da igra igru poput goa,

I'd have it play thousands of games of Go,

morao bih da odigram na hiljade partija goa,

but in the process I also teach it

ali bih ga usput naučio

how to discern a good game from a bad game.

kako da razlikuje dobru od loše partije.

If I want to create an artificially intelligent legal assistant,

Ako bih želeo da stvorim pravnog asistenta sa veštačkom inteligencijom,

I will teach it some corpus of law

podučio bih ga zakonskim pravilncima,

but at the same time I am fusing with it

ali bih mu istovremeno s tim mešao

the sense of mercy and justice that is part of that law.

osećanja milosti i pravde koja su delovi tog zakona.

In scientific terms, this is what we call ground truth,

U naučnom smislu, ovo je nešto što nazivamo temeljnom istinom,

and here's the important point:

a ovo je važna tačka:

in producing these machines,

stvarajući ove mašine,

we are therefore teaching them a sense of our values.

sledi da ih učimo svesti o našim vrednostima.

To that end, I trust an artificial intelligence

U tom smislu, verujem veštačkoj inteligenciji

the same, if not more, as a human who is well-trained.

isto, ako ne i više, kao i dobro obučenim ljudima.

But, you may ask,

Međutim, možda se pitate,

what about rogue agents,

šta je sa odmetnicima,

some well-funded nongovernment organization?

nekim dobro finansiranim nevladinim organizacijama?

I do not fear an artificial intelligence in the hand of a lone wolf.

Ne plašim se veštačke inteligencije u rukama vukova samotnjaka.

Clearly, we cannot protect ourselves against all random acts of violence,

Očito se ne možemo zaštititi od svih nasumičnih činova nasilja,

but the reality is such a system

ali u stvarnosti sličan sistem

requires substantial training and subtle training

zahteva značajanu i prefinjenu obuku

far beyond the resources of an individual.

koja prevazilazi resurse pojedinaca.

And furthermore,

A zatim,

it's far more than just injecting an internet virus to the world,

to je daleko više od pukog ubrizgavanja internet virusa u svet,

where you push a button, all of a sudden it's in a million places

gde pritisnete dugme i iznenada se nalazi na milion mesta

and laptops start blowing up all over the place.

i laptopovi počinju svuda da eksplodiraju.

Now, these kinds of substances are much larger,

Sad, ovakve materije su daleko veće,

and we'll certainly see them coming.

i sigurno ćemo ih doživeti.

Do I fear that such an artificial intelligence

Da li me je strah da će takva veštačka inteligencija

might threaten all of humanity?

možda ugroziti čitavo čovečanstvo?

If you look at movies such as "The Matrix," "Metropolis,"

Ako pogledate filmove, poput "Matriksa", "Metropolisa",

"The Terminator," shows such as "Westworld,"

"Terminatora", serije poput "Zapadnog sveta",

they all speak of this kind of fear.

svi se oni bave tim strahom.

Indeed, in the book "Superintelligence" by the philosopher Nick Bostrom,

Zaista, u knjizi "Superinteligencija" filozofa Nika Bostroma,

he picks up on this theme

on se bavi ovom tematikom

and observes that a superintelligence might not only be dangerous,

i razmatra da bi superinteligencija mogla ne samo da bude opasna,

it could represent an existential threat to all of humanity.

već bi mogla predstavljati egzistencijalnu pretnju čitavom čovečanstvu.

Dr. Bostrom's basic argument

Osnovni argument dr Bostroma

is that such systems will eventually

je da bi slični sistemi vremenom

have such an insatiable thirst for information

imali takavu nezasitu žudnju za informacijama

that they will perhaps learn how to learn

da bi možda naučili kako da uče

and eventually discover that they may have goals

i vremenom bi otkrili da možda imaju ciljeve

that are contrary to human needs.

koji su suprotni ljudskim potrebama.

Dr. Bostrom has a number of followers.

Dr Bostrom ima brojne pratioce.

He is supported by people such as Elon Musk and Stephen Hawking.

Podržavaju ga ljudi poput Ilona Maska i Stivena Hokinga.

With all due respect

Uz dužno poštovanje

to these brilliant minds,

tim briljantnim umovima,

I believe that they are fundamentally wrong.

verujem da suštinski greše.

Now, there are a lot of pieces of Dr. Bostrom's argument to unpack,

Sad, argument dr Bostroma je suviše složen za razlaganje,

and I don't have time to unpack them all,

i ja nemam vremena da ga u potpunosti razložim,

but very briefly, consider this:

ali veoma kratko, razmotrite sledeće:

super knowing is very different than super doing.

superznanje je veoma različito od superdelanja.

HAL was a threat to the Discovery crew

HAL je bio pretnja za posadu Diskaverija,

only insofar as HAL commanded all aspects of the Discovery.

samo zato što je HAL zapovedao svim aspektima Diskaverija.

So it would have to be with a superintelligence.

Dakle, moralo bi da se radi o superinteligenciji.

It would have to have dominion over all of our world.

Morala bi da ima vlast u čitavom našem svetu.

This is the stuff of Skynet from the movie "The Terminator"

Tako je u slučaju Skajneta iz filma "Terminator"

in which we had a superintelligence

u kom smo imali superinteligenciju

that commanded human will,

koja je zapovedala ljudskom voljom,

that directed every device that was in every corner of the world.

koja je upravljala svakim uređajem u svakom kutku sveta.

Practically speaking,

Prosto govoreći,

it ain't gonna happen.

to se neće desiti.

We are not building AIs that control the weather,

Ne pravimo VI koja kontroliše vremenske prilike,

that direct the tides,

koja upravlja plimama,

that command us capricious, chaotic humans.

koja zapoveda nama, hirovitim, haotičnim ljudima.

And furthermore, if such an artificial intelligence existed,

Potom, kad bi postojala slična veštačka inteligencija,

it would have to compete with human economies,

morala bi da se takmiči sa ljudskim ekonomijama,

and thereby compete for resources with us.

a time i sa nama zbog resursa.

And in the end --

I na kraju -

don't tell Siri this --

nemojte ovo reći Siri -

we can always unplug them.

možemo uvek da ih isključimo.

(Laughter)

(Smeh)

We are on an incredible journey

Na neverovatnom smo putu

of coevolution with our machines.

koevolucije sa našim mašinama.

The humans we are today

Ljudi kakvi smo danas,

are not the humans we will be then.

nećemo biti ljudi sutrašnjice.

To worry now about the rise of a superintelligence

Briga zbog uspona superinteligencije

is in many ways a dangerous distraction

je na razne načine opasno odvraćanje pažnje

because the rise of computing itself

jer uspon samog računarstva

brings to us a number of human and societal issues

nam donosi brojna ljudska i društvena pitanja

to which we must now attend.

kojima se odmah moramo baviti.

How shall I best organize society

Kako da organizujemo društvo na najbolji način

when the need for human labor diminishes?

kada se smanji potreba za ljudskom radnom snagom?

How can I bring understanding and education throughout the globe

Kako da donesemo razumevanje i obrazovanje širom sveta,

and still respect our differences?

a da i dalje poštujemo naše razlike?

How might I extend and enhance human life through cognitive healthcare?

Kako bismo mogli da proširimo i unapredimo ljudski život kroz kognitivno zdravstvo?

How might I use computing

Kako da koristimo računare

to help take us to the stars?

da nas odvedu do zvezda?

And that's the exciting thing.

A to je uzbudljivo.

The opportunities to use computing

Šansa da se koriste računari

to advance the human experience

da bismo unapredili ljudsko iskustvo

are within our reach,

su nam na dohvat ruke,

here and now,

ovde i sad,

and we are just beginning.

a tek smo počeli.

Thank you very much.

Mnogo vam hvala.

(Applause)

(Aplauz)