How will AI change the world?

Prevodilac: Milenka Okuka Lektor: Una Kavazović

U godinama koje slede, veštačka inteligencija

In the coming years, artificial intelligence

će vam verovatno promeniti život, a sasvim izvesno i čitav svet.

is probably going to change your life, and likely the entire world.

Međutim, ljudi se baš i ne slažu tačno na koji način.

But people have a hard time agreeing on exactly how.

Slede isečci iz intervjua sa Svetskog ekonomskog foruma

The following are excerpts from an interview

where renowned computer science professor and AI expert Stuart Russell

u kojima cenjeni profesor računarske nauke i ekspert za VI Stjuart Rasel

helps separate the sense from the nonsense.

pomaže da odvojimo smisao od besmislica.

There’s a big difference between asking a human to do something

Velika je razlika između traženja od ljudskog bića da uradi nešto

and giving that as the objective to an AI system.

i navođenja istog cilja sistemu VI.

When you ask a human to get you a cup of coffee,

Kada tražite od ljudskog bića da vam donese šolju kafe,

you don’t mean this should be their life’s mission,

ne smatrate da bi to trebalo da bude njihova životna misija

and nothing else in the universe matters.

i da je samo to bitno u univerzumu.

Even if they have to kill everybody else in Starbucks

Čak i ako moraju da ubiju svakoga u kafeu

to get you the coffee before it closes— they should do that.

da vam donesu kafu pre zatvaranja - trebalo bi to da urade.

No, that’s not what you mean.

Ne, ne mislite tako.

All the other things that we mutually care about,

Sve ostalo do čega nam je međusobno stalo

they should factor into your behavior as well.

trebalo bi takođe da utiče na vaše ponašanje.

And the problem with the way we build AI systems now

A problem s tim kako trenutno gradimo sisteme VI

is we give them a fixed objective.

je to što im dajemo fiksiran cilj.

The algorithms require us to specify everything in the objective.

Algoritmi od nas zahtevaju da sve definišemo u cilju.

And if you say, can we fix the acidification of the oceans?

A ako kažete, možemo li da popravimo zakiseljavanje okeana?

Yeah, you could have a catalytic reaction that does that extremely efficiently,

Da, možete da imate katalitičku reakciju koja to izuzetno uspešno postiže,

but it consumes a quarter of the oxygen in the atmosphere,

ali bi potrošila četvrtinu kiseonika u atmosferi,

which would apparently cause us to die fairly slowly and unpleasantly

zbog čega bismo očito umirali krajnje sporo i neprijatno

over the course of several hours.

tokom nekoliko sati.

So, how do we avoid this problem?

Dakle, kako da izbegnemo ovaj problem?

You might say, okay, well, just be more careful about specifying the objective—

Mogli biste reći, u redu, samo budite pažljiviji kod definisanja cilja -

don’t forget the atmospheric oxygen.

ne zaboravite atmosferski kiseonik.

And then, of course, some side effect of the reaction in the ocean

A onda, naravno, neka nuspojava reakcije u okeanu

poisons all the fish.

potruje svu ribu.

Okay, well I meant don’t kill the fish either.

U redu, mislio sam nemoj ni ribu da ubiješ.

And then, well, what about the seaweed?

A onda, dakle, šta je sa morskom travom?

Don’t do anything that’s going to cause all the seaweed to die.

Nemoj da uradiš ništa što bi uzrokovalo izumiranje morske trave.

And on and on and on.

I tako dalje, i tako dalje.

And the reason that we don’t have to do that with humans is that

A razlog zašto ne moramo to da radimo s ljudima

humans often know that they don’t know all the things that we care about.

je što ljudi često znaju da ne znaju sve stvari do kojih nam je stalo.

If you ask a human to get you a cup of coffee,

Ako zatražite od ljudskog bića da vam donese šolju kafe,

and you happen to be in the Hotel George Sand in Paris,

a nalazite se u hotelu Žorž Sand u Parizu

where the coffee is 13 euros a cup,

gde šolja kafe košta 13 evra,

it’s entirely reasonable to come back and say, well, it’s 13 euros,

sasvim je razumno da se vratite i kažete: „Košta 13 evra,

are you sure you want it, or I could go next door and get one?

jesi li siguran da je želiš ili da je kupim u susednoj radnji?”

And it’s a perfectly normal thing for a person to do.

I to je sasvim normalno za nekoga da uradi.

To ask, I’m going to repaint your house—

Da pita: „Prekrečiću ti kuću -

is it okay if I take off the drainpipes and then put them back?

je li u redu da skinem oluke i da ih potom vratim?”

We don't think of this as a terribly sophisticated capability,

Ovo ne smatramo naročito prefinjenom sposobnošću,

but AI systems don’t have it because the way we build them now,

ali sistemi VI je nemaju, jer zbog toga kako ih trenutno gradimo

they have to know the full objective.

moraju u potpunosti da znaju cilj.

If we build systems that know that they don’t know what the objective is,

Ako bismo gradili sisteme koji znaju da ne znaju šta je cilj,

then they start to exhibit these behaviors,

onda bi počeli da ispoljavaju ova ponašanja,

like asking permission before getting rid of all the oxygen in the atmosphere.

poput traženja dozvole pre nego se reše celokupnog kiseonika u atmosferi.

In all these senses, control over the AI system

U svakom smislu, kontrola nad sistemom VI

comes from the machine’s uncertainty about what the true objective is.

dolazi iz nesigurnosti mašine po pitanju njenog istinskog cilja.

And it’s when you build machines that believe with certainty

A kada gradite mašine koje veruju sa sigurnošću

that they have the objective,

da imaju cilj,

that’s when you get this sort of psychopathic behavior.

tada dobijate psihopatsko ponašanje.

And I think we see the same thing in humans.

A smatram da isto to vidimo i kod ljudi.

What happens when general purpose AI hits the real economy?

Šta se desi kada veštačka opšta inteligencija stigne u realnu ekonomiju?

How do things change? Can we adapt?

Kako se stvari menjaju? Možemo li se prilagoditi?

This is a very old point.

Radi se o izuzetno staroj poenti.

Amazingly, Aristotle actually has a passage where he says,

Izvanredno je da Aristotel zapravo ima pasus gde kaže,

look, if we had fully automated weaving machines

vidite, kad bismo imali potpuno automatizovane razboje

and plectrums that could pluck the lyre and produce music without any humans,

i trzalice koje bi mogle da trzaju liru i proizvode muziku bez ljudi,

then we wouldn’t need any workers.

onda nam ne bi bili potrebni radnici.

That idea, which I think it was Keynes

Ta zamisao, koja mislim da je bila Kejnsova

who called it technological unemployment in 1930,

koji ju je 1930. godine nazvao tehnološkom nezaposlenošću

is very obvious to people.

je krajnje očita ljudima.

They think, yeah, of course, if the machine does the work,

Misle, da, naravno, ako mašina obavlja posao,

then I'm going to be unemployed.

onda ću biti nezaposlen.

You can think about the warehouses that companies are currently operating

Setite se skladišta kojima trenutno upravljaju kompanije

for e-commerce, they are half automated.

za e-trgovinu, ona su poluautomatizovana.

The way it works is that an old warehouse— where you’ve got tons of stuff piled up

Način na koji to funkcioniše je da stari magacin - gde imate tone stvari u hrpama

all over the place and humans go and rummage around

svuda razbacane i ljudi dolaze i preturaju okolo

and then bring it back and send it off—

i potom to donose i odašiljaju -

there’s a robot who goes and gets the shelving unit

postoji robot koji odlazi i uzima paletni regal

that contains the thing that you need,

u kom se nalazi stvar koja vam je potrebna,

but the human has to pick the object out of the bin or off the shelf,

ali čovek mora da uzme predmet iz kontejnera ili sa police

because that’s still too difficult.

jer je to i dalje suviše komplikovano.

But, at the same time,

Međutim, istovremeno,

would you make a robot that is accurate enough to be able to pick

da li biste napravili robota koji je dovoljno precizan da može da odabere

pretty much any object within a very wide variety of objects that you can buy?

gotovo svaki predmet u širokom dijapazonu predmeta koje možete da kupite?

That would, at a stroke, eliminate 3 or 4 million jobs?

Nešto što bi, jednim potezom, eliminisalo tri ili četiri miliona poslova?

There's an interesting story that E.M. Forster wrote,

Postoji zanimljiva priča koju je napisao E. M. Forster

where everyone is entirely machine dependent.

u kojoj je svako u potpunosti zavisan od mašina.

The story is really about the fact that if you hand over

Priča zaista govori o činjenici da ako predate

the management of your civilization to machines,

upravljanje civilizacijom mašinama,

you then lose the incentive to understand it yourself

gubite podsticaj da ih sami razumete

or to teach the next generation how to understand it.

ili da podučite narednu generaciju kako da ih razumeju.

You can see “WALL-E” actually as a modern version,

Možete posmatrati „Volija” zapravo kao savremenu verziju,

where everyone is enfeebled and infantilized by the machine,

gde su mašine svakoga oslabile i infantilizovale,

and that hasn’t been possible up to now.

a to nije bilo moguće sve do sad.

We put a lot of our civilization into books,

Veći deo naše civilizacije smo smestili u knjige,

but the books can’t run it for us.

ali one ne mogu da upravljaju umesto nas.

And so we always have to teach the next generation.

Stoga, uvek moramo da podučavamo sledeću generaciju.

If you work it out, it’s about a trillion person years of teaching and learning

Ako to izračunate, radi se o oko bilion životnih dobi podučavanja i učenja

and an unbroken chain that goes back tens of thousands of generations.

i neprekidnom lancu koji ide unazad desetinama hiljada generacija.

What happens if that chain breaks?

Šta se desi kada se lanac prekine?

I think that’s something we have to understand as AI moves forward.

Mislim da je to nešto što moramo da razumemo kako VI napreduje.

The actual date of arrival of general purpose AI—

Stvarni datum dolaska veštačke opšte inteligencije -

you’re not going to be able to pinpoint, it isn’t a single day.

ne biste mogli tačno da odredite, ne radi se o jednom danu.

It’s also not the case that it’s all or nothing.

Takođe nije slučaj gde je sve ili ništa.

The impact is going to be increasing.

Njen uticaj će da raste.

So with every advance in AI,

Stoga sa svakim unapređenjem VI,

it significantly expands the range of tasks.

ona značajno proširuje dijapazon zadataka.

So in that sense, I think most experts say by the end of the century,

Te u tom smislu, smatram da većina stručnjaka kaže da do kraja veka,

we’re very, very likely to have general purpose AI.

veoma, veoma je verovatno da ćemo imati veštačku opštu inteligenciju.

The median is something around 2045.

Srednja vrednost je oko 2045. godine.

I'm a little more on the conservative side.

Naginjem više ka konzervativnoj strani.

I think the problem is harder than we think.

Smatram da je problem teži nego što mislimo.

I like what John McAfee, he was one of the founders of AI,

Sviđa mi se ono što je Džon Makafi, bio je jedan od osnivača VI,

when he was asked this question, he said, somewhere between five and 500 years.

kada su ga upitali o ovome rekao, negde između pet i 500 godina.

And we're going to need, I think, several Einsteins to make it happen.

I biće nam potrebno, mislim, nekoliko Ajnštajna da to i ostvarimo.