We're building a dystopia just to make people click on ads | Zeynep Tufekci

Prevodilac: Ivana Korom Lektor: Aleksandar Korom

So when people voice fears of artificial intelligence,

Kada ljudi pričaju o strahu od veštačke inteligencije,

very often, they invoke images of humanoid robots run amok.

veoma često se pozivaju na slike humanoidnih robota koji divljaju.

You know? Terminator?

Znate? Terminator?

You know, that might be something to consider,

Znate, o tome bi možda i trebalo da razmišljamo,

but that's a distant threat.

ali to je daleka pretnja.

Or, we fret about digital surveillance

Ili se plašimo digitalnog nadzora

with metaphors from the past.

putem metafora iz prošlosti.

"1984," George Orwell's "1984,"

"1984" Džordža Orvela

it's hitting the bestseller lists again.

ponovo se našla na listi bestselera.

It's a great book,

To je odlična knjiga,

but it's not the correct dystopia for the 21st century.

ali ne i tačan opis distopije za 21. vek.

What we need to fear most

Ono čega bi najviše trebalo da se plašimo

is not what artificial intelligence will do to us on its own,

nije ono što veštačka inteligencija može da nam uradi sama od sebe,

but how the people in power will use artificial intelligence

nego kako će moćnici koristiti veštačku inteligenciju

to control us and to manipulate us

kako bi nas kontrolisali i manipulisali nama

in novel, sometimes hidden,

na neki nov, ponekad diskretan,

subtle and unexpected ways.

suptilan i neočekivan način.

Much of the technology

Tehnologije koje će u skorijoj budućnosti

that threatens our freedom and our dignity in the near-term future

početi da prete našoj slobodi i našem dostojanstvu

is being developed by companies

većinom razvijaju velike kompanije

in the business of capturing and selling our data and our attention

koje se bave prikupljanjem i prodajom naših informacija i naše pažnje

to advertisers and others:

oglašivačima i drugima:

Facebook, Google, Amazon,

Fejsbuk, Gugl, Amazon,

Alibaba, Tencent.

Alibaba, Tencent.

Now, artificial intelligence has started bolstering their business as well.

Sad, i veštačka inteligencija je postala potpora njihovog poslovanja.

And it may seem like artificial intelligence

I iako se možda čini da će veštačka inteligencija

is just the next thing after online ads.

postati trend odmah posle internet reklama,

It's not.

nije tako.

It's a jump in category.

To nije ista kategorija.

It's a whole different world,

To je potpuno drugi svet

and it has great potential.

koji ima veliki potencijal.

It could accelerate our understanding of many areas of study and research.

To može ubrzati naše razumevanje mnogih naučnih oblasti i istraživanja.

But to paraphrase a famous Hollywood philosopher,

Ali da parafraziram poznatog holivudkog filozofa:

"With prodigious potential comes prodigious risk."

"Ogroman potencijal prati i ogroman rizik."

Now let's look at a basic fact of our digital lives, online ads.

Hajde da se vratimo na osnovu naših digitalnih života - internet reklame.

Right? We kind of dismiss them.

Mi ih nekako odbacujemo, zar ne?

They seem crude, ineffective.

Čine nam se banalnim, uzaludnim.

We've all had the experience of being followed on the web

Svi smo doživeli da se osećamo kao da nas neko posmatra na internetu

by an ad based on something we searched or read.

zahvaljujući reklami baziranoj na onome što smo tražili ili pročitali.

You know, you look up a pair of boots

Znate kako ide, jedan dan tražite čizme

and for a week, those boots are following you around everywhere you go.

i narednih neoliko dana te čizme vas prate kuda god da odete.

Even after you succumb and buy them, they're still following you around.

Čak i nakon što se povinujete i kupite ih, one vas i dalje prate svuda okolo.

We're kind of inured to that kind of basic, cheap manipulation.

Mi smo nekako uvučeni u taj tip osnovne, jeftine manipulacije.

We roll our eyes and we think, "You know what? These things don't work."

Zakolutamo očima i pomislimo: "Znate šta?

Ovako nešto ne može da funkcioniše."

Except, online,

Osim što na internetu digitalne tehnologije nisu samo reklame.

the digital technologies are not just ads.

Now, to understand that, let's think of a physical world example.

Da bismo to razumeli, uzmimo primer iz fizičkog sveta.

You know how, at the checkout counters at supermarkets, near the cashier,

Znate kako blizu kase u supermarketu

there's candy and gum at the eye level of kids?

slatkiši i žvake uvek stoje u nivou očiju male dece?

That's designed to make them whine at their parents

To je dizajnirano da bi deca počela da se prenemažu

just as the parents are about to sort of check out.

baš kada roditelji treba da plate račun.

Now, that's a persuasion architecture.

E to vam je arhitektura ubeđivanja.

It's not nice, but it kind of works.

Nije lepo, ali funkcioniše.

That's why you see it in every supermarket.

Zato to vidite u svakom supermarketu.

Now, in the physical world,

Sada, u fizičkom svetu,

such persuasion architectures are kind of limited,

takva arhitektura ubeđivanja ima svoje granice,

because you can only put so many things by the cashier. Right?

jer pored kase možete da stavite samo određen broj stvari, zar ne?

And the candy and gum, it's the same for everyone,

A slatkiši i žvake isti su za svakoga

even though it mostly works

iako ovo obično funkcioniše

only for people who have whiny little humans beside them.

samo kod ljudi koji vode male kenjkave osobe sa sobom.

In the physical world, we live with those limitations.

U fizičkom svetu živimo sa ovakvim ograničenjima.

In the digital world, though,

U digitalnom svetu, doduše,

persuasion architectures can be built at the scale of billions

mogu da se izgrade milijarde objekata po arhitekturi ubeđivanja

and they can target, infer, understand

i oni mogu da targetiraju, zaključuju, razumeju

and be deployed at individuals

i mogu da budu raspoređeni prema pojedincima

one by one

jedan po jedan

by figuring out your weaknesses,

tako što će razumevati naše slabosti,

and they can be sent to everyone's phone private screen,

i mogu da se pošalju na svačiji privatni ekran

so it's not visible to us.

tako da ne budu svima vidljivi.

And that's different.

I tu leži ta razlika.

And that's just one of the basic things that artificial intelligence can do.

To je samo jedna od osnovnih stvari koje veštačka inteligencija može da uradi.

Now, let's take an example.

Sada, uzmimo jedan primer.

Let's say you want to sell plane tickets to Vegas. Right?

Recimo da želite da prodate avionske karte za Vegas.

So in the old world, you could think of some demographics to target

U starom svetu mogli biste se setiti neke demografije koju treba da targetirate

based on experience and what you can guess.

na osnovu iskustva i onoga što možete da pogodite.

You might try to advertise to, oh,

Mogli biste da probate da ih reklamirate,

men between the ages of 25 and 35,

recimo, muškarcima od 25 do 35 godina

or people who have a high limit on their credit card,

ili ljudima koji imaju visok limit na svojim kreditnim karticama,

or retired couples. Right?

ili parovima u penziji. Zar ne?

That's what you would do in the past.

To biste uradili u prošlosti.

With big data and machine learning,

Zahvaljujući velikoj količini podataka i mašinskom učenju,

that's not how it works anymore.

to tako više ne funkcioniše.

So to imagine that,

Da biste to sada zamislili,

think of all the data that Facebook has on you:

setite se svih onih podataka o vama koje poseduje Fejsbuk:

every status update you ever typed,

svaki status koji ste ikada napisali,

every Messenger conversation,

svaki razgovor na Mesindžeru,

every place you logged in from,

svako mesto sa kojeg ste se ulogovali,

all your photographs that you uploaded there.

sve vaše fotografije koje ste tamo postavili.

If you start typing something and change your mind and delete it,

Čak i kada počnete da pišete nešto pa se predomislite i obrišete,

Facebook keeps those and analyzes them, too.

Fejsbuk čuva te informacije i takođe ih analizira.

Increasingly, it tries to match you with your offline data.

Sve više pokušava da vas poveže sa informacijama koje nisu na internetu.

It also purchases a lot of data from data brokers.

Uz to i kupuje veliki broj informacija od data brokera.

It could be everything from your financial records

To može biti sve, od vaših finansijskih izveštaja,

to a good chunk of your browsing history.

sve do dobrog dela istorije u vašem pretraživaču.

Right? In the US, such data is routinely collected,

U SAD se takve informacije rutinski

prikupljaju, sređuju i prodaju.

collated and sold.

In Europe, they have tougher rules.

U Evropi postoje čvršća pravila.

So what happens then is,

Tako se dešava da mešanjem svih tih informacija

by churning through all that data, these machine-learning algorithms --

ovi algoritmi mašinskog učenja -

that's why they're called learning algorithms --

jer tako se zovu: algoritmi za učenje -

they learn to understand the characteristics of people

oni pokušavaju da razumeju karakteristike ljudi

who purchased tickets to Vegas before.

koji su ranije kupili karte za Vegas.

When they learn this from existing data,

Kada ovo nauče od postojećih informacija

they also learn how to apply this to new people.

takođe nauče kako isto da primene na drugim ljudima.

So if they're presented with a new person,

Pa ako im se predstavi nova osoba, mogu da je svrstaju među one

they can classify whether that person is likely to buy a ticket to Vegas or not.

koje će verovatno kupiti karte za Vegas ili među one koji neće.

Fine. You're thinking, an offer to buy tickets to Vegas.

Dobro. Vi sada mislite: "Ponuda za karte za Vegas.

I can ignore that.

Ignorisaću to."

But the problem isn't that.

Ali problem ne leži u tome.

The problem is,

Problem je što mi više ne razumemo

we no longer really understand how these complex algorithms work.

kako zaista ti algoritmi funkcionišu.

We don't understand how they're doing this categorization.

Ne razumemo kako oni prave tu kategorizaciju.

It's giant matrices, thousands of rows and columns,

Njihove grandiozne metrike, hiljade redova i kolona,

maybe millions of rows and columns,

možda milione redova i kolona,

and not the programmers

a ne čak ni programeri,

and not anybody who looks at it,

ili bilo ko drugi ko radi na tome,

even if you have all the data,

čak i ako poseduju sve informacije

understands anymore how exactly it's operating

ne razumeju kako tačno oni rade

any more than you'd know what I was thinking right now

ništa više od onoga što vi znate o onome o čemu ja trenutno razmišljam

if you were shown a cross section of my brain.

čak i ako biste videli poprečni presek mog mozga.

It's like we're not programming anymore,

Čini se da mi više ne programiramo,

we're growing intelligence that we don't truly understand.

mi gajimo inteligenciju koju istinski ne razumemo.

And these things only work if there's an enormous amount of data,

A ove stvari funkcionišu samo ako postoji ogromna količina podataka,

so they also encourage deep surveillance on all of us

tako da oni takođe ohrabruju teški nadzor nad svima nama

so that the machine learning algorithms can work.

kako bi algoritmi mašinskog učenja funkcionisali.

That's why Facebook wants to collect all the data it can about you.

Zbog toga Fejsbuk želi da prikupi sve moguće informacije o vama.

The algorithms work better.

Da bi algoritmi radili bolje.

So let's push that Vegas example a bit.

Hajde da se još malo poigramo sa primerom o Vegasu.

What if the system that we do not understand

Šta ako je sistem koji ne razumemo

was picking up that it's easier to sell Vegas tickets

shvatio da je lakše prodati karte za Vegas

to people who are bipolar and about to enter the manic phase.

ljudima koji su bipolarni i koji su na pragu manične faze.

Such people tend to become overspenders, compulsive gamblers.

Ovakvi ljudi prelako troše novac i imaju tendenciju da postanu kompulsivni kockari.

They could do this, and you'd have no clue that's what they were picking up on.

Oni to mogu da urade, a da vi nemate pojma da su baš te informacije prikupljali.

I gave this example to a bunch of computer scientists once

Jednom sam dala ovaj primer grupi kompjuterskih naučnika

and afterwards, one of them came up to me.

da bi mi, nakon toga, jedan od njih prišao.

He was troubled and he said, "That's why I couldn't publish it."

Nešto ga je mučilo i rekao je: "Zato nisam mogao da ga objavim."

I was like, "Couldn't publish what?"

A ja sam ga pitala: "Šta niste mogli da objavite?"

He had tried to see whether you can indeed figure out the onset of mania

Pokušao je da ustanovi da li zatista može da se utvrdi početak u manične faze

from social media posts before clinical symptoms,

na osnovu postova na društvenim mrežama pre nego što se pojave klinički simptomi,

and it had worked,

i uspelo mu je,

and it had worked very well,

veoma mu je uspelo,

and he had no idea how it worked or what it was picking up on.

a da on nije imao pojma kako je uspelo i kakve informacije su se prikupljale.

Now, the problem isn't solved if he doesn't publish it,

No, problem se neće rešiti ukoliko on to ne objavi,

because there are already companies

jer već postoje kompanije

that are developing this kind of technology,

koje razvijaju ovakvu tehnologiju,

and a lot of the stuff is just off the shelf.

i mnogo je stvari sada dostupno.

This is not very difficult anymore.

To više nije tako komplikovano.

Do you ever go on YouTube meaning to watch one video

Da li vam se ikada desilo da odete na Jutjub da pogledate jedan video

and an hour later you've watched 27?

i sat vremena kasnije pogledali ste već njih 27?

You know how YouTube has this column on the right

Znate kako Jutjub ima onu kolonu sa desne strane

that says, "Up next"

koja kaže "Sledeće..."

and it autoplays something?

pa onda automatski pusti nešto?

It's an algorithm

To je algoritam

picking what it thinks that you might be interested in

koji pokazuje ono za šta misli da bi moglo da vas interesuje,

and maybe not find on your own.

a da možda sami nećete naći.

It's not a human editor.

To nije živi urednik.

It's what algorithms do.

To rade algoritmi.

It picks up on what you have watched and what people like you have watched,

Oni prikupljaju sve ono što ste gledali i ono što su gledali ljudi poput vas,

and infers that that must be what you're interested in,

i zaključuje da bi to trebalo da bude ono što vas interesuje,

what you want more of,

ono čega želite još,

and just shows you more.

pa vam pokazuje još toga.

It sounds like a benign and useful feature,

To deluje kao benigna i korisna odlika,

except when it isn't.

ali ipak nije.

So in 2016, I attended rallies of then-candidate Donald Trump

Tokom 2016. sam posećivala mitinge tadašnjeg kandidata Donalda Trampa

to study as a scholar the movement supporting him.

kako bih kao naučnik izučavala pokrete koji ga podržavaju.

I study social movements, so I was studying it, too.

Ja se bavim društvenim pokretima, tako da sam i njih izučavala.

And then I wanted to write something about one of his rallies,

I onda sam htela da napišem nešto o jednom od njegovih mitinga,

so I watched it a few times on YouTube.

pa sam ga nekoliko puta pogledala na Jutjubu.

YouTube started recommending to me

Jutjub je onda počeo da mi preporučuje

and autoplaying to me white supremacist videos

i automatski pušta video snimke belačkih supremacističkih grupa

in increasing order of extremism.

po rastućem redosledu ekstremizma.

If I watched one,

Ako bih pogledala jedan,

it served up one even more extreme

spremio bi mi jedan još ekstremniji

and autoplayed that one, too.

pa bi mi ga pustio automatski, takođe.

If you watch Hillary Clinton or Bernie Sanders content,

Ako biste gledali sadržaj vezan za Hilari Klinton ili Bernija Sandersa,

YouTube recommends and autoplays conspiracy left,

Jutjub bi vam preporučio i automatski puštao snimke o zaveri levice,

and it goes downhill from there.

i od toga sve odlazi nizbrdo.

Well, you might be thinking, this is politics, but it's not.

Pa, mogli biste pomisliti, ovo je politika, ali nije.

This isn't about politics.

To nema veze sa politikom.

This is just the algorithm figuring out human behavior.

To je samo algoritam u pokušaju da prokljuvi ljudsko ponašanje.

I once watched a video about vegetarianism on YouTube

Jednom sam na Jutjubu gledala video o vegetarijanstvu,

and YouTube recommended and autoplayed a video about being vegan.

pa mi je Jutjub preporučio i automatski pustio video o tome kako je biti vegan.

It's like you're never hardcore enough for YouTube.

Kao da ništa nije dovoljno hardkor za Jutjub.

(Laughter)

(Smeh)

So what's going on?

Pa, o čemu se radi?

Now, YouTube's algorithm is proprietary,

Jutjubov algoritam je vlasnički,

but here's what I think is going on.

ali evo šta ja mislim o tome.

The algorithm has figured out

Algoritam je shvatio da

that if you can entice people

ako možete da namamite ljude

into thinking that you can show them something more hardcore,

da misle da možete da im pokažete nešto još više hardkor

they're more likely to stay on the site

verovatnije je da će ostati na sajtu

watching video after video going down that rabbit hole

i gledati video za videom, kao da propadaju kroz zečiju rupu,

while Google serves them ads.

dok im Gugl servira reklame.

Now, with nobody minding the ethics of the store,

Sad, dok god nikoga ne brine etika prodavnice,

these sites can profile people

ovi sajtovi mogu da profilišu ljude

who are Jew haters,

koji mrze Jevreje,

who think that Jews are parasites

koji misle da su Jevreji paraziti,

and who have such explicit anti-Semitic content,

i koji poseduju tako eksplicitni antisemitski sadržaj

and let you target them with ads.

i dopuštaju vam da ih vaše reklame targetiraju.

They can also mobilize algorithms

Oni takođe mogu da mobilizuju algoritme

to find for you look-alike audiences,

ne bi li vam našli slične publike,

people who do not have such explicit anti-Semitic content on their profile

ljude koji ne poseduju tako eksplicitni antisemitski sadržaj na svojim profilima

but who the algorithm detects may be susceptible to such messages,

ali koje algoritam prepoznaje kao podložne takvim porukama,

and lets you target them with ads, too.

i dopušta vam da i njih targetirate.

Now, this may sound like an implausible example,

Sad, ovo će možda zvučati kao malo verovatan primer,

but this is real.

ali je realan.

ProPublica investigated this

ProPublika je istraživala ovo

and found that you can indeed do this on Facebook,

i saznala da to zaista možete da uradite na Fejsbuku

and Facebook helpfully offered up suggestions

i da će vam Fejsbuk ljubazno ponuditi još predloga

on how to broaden that audience.

kako da proširite auditorijum.

BuzzFeed tried it for Google, and very quickly they found,

Bazfid je isto pokušao sa Guglom i veoma brzo su shvatili da,

yep, you can do it on Google, too.

naravno, isto možete da uradite i na Guglu.

And it wasn't even expensive.

A nije bilo ni skupo.

The ProPublica reporter spent about 30 dollars

Reporter ProPublike potrošio je oko 30 dolara

to target this category.

na targetiranje ove kategorije.

So last year, Donald Trump's social media manager disclosed

Tako je prošle godine menadžer društvenih stranica Donalda Trampa

that they were using Facebook dark posts to demobilize people,

otkrio da su koristili Fejsbukove mračne postove da demobilišu ljude,

not to persuade them,

ne da ih navedu,

but to convince them not to vote at all.

već da ih ubede da ne glasaju uopšte.

And to do that, they targeted specifically,

A da bi to uradili, targetirali su posebno,

for example, African-American men in key cities like Philadelphia,

na primer, muškarce Afroamerikance iz ključnih gradova poput Filadelfije,

and I'm going to read exactly what he said.

i sada ću vam pročitati tačno šta je on rekao.

I'm quoting.

Citiram:

They were using "nonpublic posts

Koristili su "privatne postove

whose viewership the campaign controls

čije preglede kontroliše kampanja

so that only the people we want to see it see it.

kako bi ih videli samo oni ljudi za koje želimo da ih vide.

We modeled this.

Mi smo to modelovali.

It will dramatically affect her ability to turn these people out."

To će dramatično uticati na njenu sposobnost da odbije ove ljude."

What's in those dark posts?

Šta je u ovim mračnim postovima?

We have no idea.

Mi nemamo pojma.

Facebook won't tell us.

Fejsbuk neće da nam kaže.

So Facebook also algorithmically arranges the posts

To znači da Fejsbuk postove vaših prijatelja ili stranica koje pratite

that your friends put on Facebook, or the pages you follow.

raspoređuje po algoritmu.

It doesn't show you everything chronologically.

Ne pokazuje vam sve hronološki.

It puts the order in the way that the algorithm thinks will entice you

Raspored utvrđuje po tome šta algoritam misli da će vas namamiti

to stay on the site longer.

da što duže ostanete na sajtu.

Now, so this has a lot of consequences.

Sve to ima brojne posledice.

You may be thinking somebody is snubbing you on Facebook.

Možda mislite da vas neko ignoriše na Fejsbuku,

The algorithm may never be showing your post to them.

a zapravo im Fejsbuk nikada ne pokazuje vaše postove.

The algorithm is prioritizing some of them and burying the others.

Algoritam daje prednost nekima dok druge ukopava.

Experiments show

Eksperimenti pokazuju da

that what the algorithm picks to show you can affect your emotions.

ono što algoritam odabere da vam pokaže može da utiče na vaše emocije.

But that's not all.

Ali, to nije sve.

It also affects political behavior.

Takođe utiče na političko ponašanje.

So in 2010, in the midterm elections,

2010. godine, usred srednjoročnih izbora,

Facebook did an experiment on 61 million people in the US

Fejsbuk je napravio eksperiment nad 61 milionom ljudi u SAD

that was disclosed after the fact.

koji je kasnije obelodanjen.

So some people were shown, "Today is election day,"

Neki ljudi su videli: "Danas su izbori",

the simpler one,

jednostavnija verzija,

and some people were shown the one with that tiny tweak

dok je nekima puštena ona pomalo izmenjena,

with those little thumbnails

sa onim malim slikama

of your friends who clicked on "I voted."

svih vaših prijateljima koji su kliknuli na "Glasao sam".

This simple tweak.

Mala je razlika.

OK? So the pictures were the only change,

OK? Dakle, samo su se slike razlikovale,

and that post shown just once

i taj post pušten samo jednom

turned out an additional 340,000 voters

odbio je dodatnih 34.000 glasača

in that election,

na tim izborima,

according to this research

sudeći po ovom istraživanju

as confirmed by the voter rolls.

a što je potvrđeno glasačkim listićima.

A fluke? No.

Slučajnost? Nije.

Because in 2012, they repeated the same experiment.

Jer su 2012. ponovili isti eksperiment.

And that time,

I taj put

that civic message shown just once

ta građanska poruka viđena samo jednom

turned out an additional 270,000 voters.

odbila je dodatnih 270.000 glasača.

For reference, the 2016 US presidential election

Radi poređenja, rezultat predsedničkih izbora u SAD 2016. godine

was decided by about 100,000 votes.

odredilo je oko 100.000 glasova.

Now, Facebook can also very easily infer what your politics are,

Sad, Fejsbuk takođe može da vrlo lako da zaključi koji su vaši politički stavovi

even if you've never disclosed them on the site.

čak i ako ih nikada niste otvoreno pokazali na ovom sajtu.

Right? These algorithms can do that quite easily.

Jasno? To ovi algoritmi mogu vrlo lako da urade.

What if a platform with that kind of power

Šta ako platforma koja ima toliku moć

decides to turn out supporters of one candidate over the other?

odluči da izostavi pristalice jednog kandidata u odnosu na druge?

How would we even know about it?

Kako bismo ikada saznali za to?

Now, we started from someplace seemingly innocuous --

Počeli smo nečim čini se bezazlenim -

online adds following us around --

internet reklame nas prate svuda -

and we've landed someplace else.

a mi smo otišli negde drugde.

As a public and as citizens,

Kao javnost i kao građani,

we no longer know if we're seeing the same information

mi više ne znamo da li vidimo iste informacije

or what anybody else is seeing,

niti šta bilo ko drugi vidi,

and without a common basis of information,

a bez zajedničke baze informacija,

little by little,

malo po malo,

public debate is becoming impossible,

javna debata postaje nemoguća,

and we're just at the beginning stages of this.

a mi smo još samo u početnoj fazi ovoga.

These algorithms can quite easily infer

Ovi algoritmi mogu vrlo lako da izvedu zaključak

things like your people's ethnicity,

o stvarima poput etničke pripadnosti ljudi,

religious and political views, personality traits,

religiji i političkim stavovima, ličnim osobinama,

intelligence, happiness, use of addictive substances,

inteligenciji, sreći, zavisnosti o supstancama,

parental separation, age and genders,

separaciji roditelja, polu i rodu,

just from Facebook likes.

samo po lajkovima na Fejsbuku.

These algorithms can identify protesters

Ovi algoritmi mogu identifikovati demonstrante

even if their faces are partially concealed.

čak i ako su njihova lica delimično skrivena.

These algorithms may be able to detect people's sexual orientation

Ovakvi algoritmi mogu biti sposobni da otkriju seksualnu orijentaciju ljudi

just from their dating profile pictures.

samo po njihovnim slikama na profilma za traženje partnera.

Now, these are probabilistic guesses,

Ovo su verovatne pretpostavke,

so they're not going to be 100 percent right,

što znači da neće biti 100 posto tačne,

but I don't see the powerful resisting the temptation to use these technologies

ali ne vidim moćnike kako odbijaju iskušenju da iskoriste ove tehnologije

just because there are some false positives,

samo zato što postoje neki lažni pozitivi,

which will of course create a whole other layer of problems.

koje će svakako napraviti sasvim novi sloj problema.

Imagine what a state can do

Zamislite šta država može da uradi

with the immense amount of data it has on its citizens.

sa ogromnom količinom podataka koje poseduje o svojim građanima.

China is already using face detection technology

Kina već koristi tehnologiju prepoznavanja lica

to identify and arrest people.

za identifikaciju i hapšenje ljudi.

And here's the tragedy:

U ovome je tragedija:

we're building this infrastructure of surveillance authoritarianism

pravimo infrastrukturu nadzorskog autoritarizma

merely to get people to click on ads.

samo da bismo naveli ljude da klikću na reklame.

And this won't be Orwell's authoritarianism.

A ovo neće biti orvelovski autoritatizam.

This isn't "1984."

Ovo nije "1984".

Now, if authoritarianism is using overt fear to terrorize us,

Sada, ako autoritarizam otvoreno koristi strah da bi nas terorisao,

we'll all be scared, but we'll know it,

svi bismo bili uplašeni, ali bismo ga prepoznali,

we'll hate it and we'll resist it.

mrzeli bismo ga i oduprli mu se.

But if the people in power are using these algorithms

Ali, ako ove algoritme koriste moćnici

to quietly watch us,

da nas posmatraju iz prikrajka,

to judge us and to nudge us,

da sude o nama i da nas gurkaju,

to predict and identify the troublemakers and the rebels,

da predvide i identifikuju problematične i buntovnike,

to deploy persuasion architectures at scale

da što više razviju arhitekturu ubeđivanja

and to manipulate individuals one by one

i da manipulišu pojedincima jednim za drugim

using their personal, individual weaknesses and vulnerabilities,

koristeći njihove lične slabosti i ranjivosti,

and if they're doing it at scale

a ako to rade na taj način

through our private screens

kroz naše privatne ekrane

so that we don't even know

tako da mi ni ne znamo

what our fellow citizens and neighbors are seeing,

šta naši sugrađani i komšije vide,

that authoritarianism will envelop us like a spider's web

takav autoritarizam će nas uhvatiti kao u paukovu mrežu

and we may not even know we're in it.

i možda nećemo ni znati da smo u njoj.

So Facebook's market capitalization

Dakle, Fejsbukov tržišni kapital

is approaching half a trillion dollars.

se približava polovini biliona dolara.

It's because it works great as a persuasion architecture.

To je zato što odlično funkcioniše kao arhitektura ubeđivanja.

But the structure of that architecture

Ali, struktura te arhitekture je ista

is the same whether you're selling shoes

bez obzira da li prodajete cipele

or whether you're selling politics.

ili političke stavove.

The algorithms do not know the difference.

Algoritam ne poznaje razliku.

The same algorithms set loose upon us

Isti oni algoritmi koji se koriste nad nama

to make us more pliable for ads

da bi nas učinili manje otpornim na reklame

are also organizing our political, personal and social information flows,

takođe organizuju protok političkih, ličnih i društvenih informacija do nas,

and that's what's got to change.

i to je ono što mora da se promeni.

Now, don't get me wrong,

Nemojte pogrešno da me shvatite,

we use digital platforms because they provide us with great value.

digitalne platforme koristimo jer su nam veoma dragocene.

I use Facebook to keep in touch with friends and family around the world.

Ja koristim Fejsbuk da budem u kontaktu sa prijatejima i porodicom širom sveta.

I've written about how crucial social media is for social movements.

Pisala sam o tome da su društvene mreže od ključne važnosti za društvene pokrete.

I have studied how these technologies can be used

Izučavala sam kako ove tehnologije mogu da budu korišćene

to circumvent censorship around the world.

širom sveta kako bi se izbegla cenzura.

But it's not that the people who run, you know, Facebook or Google

Ali, to ne znači da su ljudi koji vode Fejsbuk ili Gugl zli,

are maliciously and deliberately trying

niti da namerno pokušavaju da još više razdvoje

to make the country or the world more polarized

zemlju ili svet

and encourage extremism.

i da podržavaju ekstremizam.

I read the many well-intentioned statements

Pročitala sam mnogo dobronamernih

that these people put out.

izjava ovih ljudi.

But it's not the intent or the statements people in technology make that matter,

Ali nije važna namera ili izjave koje daju ljudi iz tehnologije,

it's the structures and business models they're building.

nego strukture i modeli poslovanja koje grade.

And that's the core of the problem.

To je srž problema.

Either Facebook is a giant con of half a trillion dollars

Ili je Fejsbuk ogromna prevara od pola biliona dolara

and ads don't work on the site,

i oglasi ne rade na sajtu,

it doesn't work as a persuasion architecture,

ne funkcioniše kao arhitektura ubeđivanja,

or its power of influence is of great concern.

ili je njegova moć uticaja veoma zabrinjavajuća.

It's either one or the other.

Ili jedno ili drugo.

It's similar for Google, too.

Slično je i sa Guglom.

So what can we do?

Šta možemo da uradimo?

This needs to change.

Ovo mora de se promeni.

Now, I can't offer a simple recipe,

Ne mogu da ponudim jednostavan recept,

because we need to restructure

jer moramo da restrukturišemo

the whole way our digital technology operates.

ceo način na koji digitalna tehnologija radi.

Everything from the way technology is developed

Sve od načina na koji se tehnologija razvija

to the way the incentives, economic and otherwise,

to načina na koji su motivacija, ekonomija i drugo

are built into the system.

utkani u sistem.

We have to face and try to deal with

Moramo da se suočimo i pokušamo da se izborimo

the lack of transparency created by the proprietary algorithms,

sa nedostatkom transparentnosti koji stvaraju zaštićeni algoritmi,

the structural challenge of machine learning's opacity,

strukturalnim izazovom razumljivosti mašinskog učenja,

all this indiscriminate data that's being collected about us.

svim ovim nasumičnim podacima koji se o nama prikupljaju.

We have a big task in front of us.

Pred nama je veliki zadatak.

We have to mobilize our technology,

Moramo da upregnemo našu tehnologiju,

our creativity

našu kreativnost,

and yes, our politics

i da, našu politiku

so that we can build artificial intelligence

kako bismo izgradili veštačku inteligenciju

that supports us in our human goals

koja nas podržava u našim humanim ciljevima

but that is also constrained by our human values.

ali je i ograničena našim humanim vrednostima.

And I understand this won't be easy.

Razumem da ovo neće biti lako.

We might not even easily agree on what those terms mean.

Možda se čak ni ne slažemo šta ti termini znače.

But if we take seriously

Ali ako ozbiljno shvatimo

how these systems that we depend on for so much operate,

kako funkcionišu ovi sistemi od kojih toliko zavisimo,

I don't see how we can postpone this conversation anymore.

ne vidim kako možemo da odložimo taj razgovor.

These structures

Ove strukture

are organizing how we function

organizuju naše funkcionisanje

and they're controlling

i one kontrolišu

what we can and we cannot do.

šta možemo i šta ne možemo da uradimo.

And many of these ad-financed platforms,

I mnoge od ovih platformi koje finansiraju reklame,

they boast that they're free.

one se hvale da su besplatne.

In this context, it means that we are the product that's being sold.

U ovom kontekstu, to znači da smo mi proizvod koji se prodaje.

We need a digital economy

Potrebna nam je digitalna ekonomija

where our data and our attention

gde naši podaci i naša pažnja

is not for sale to the highest-bidding authoritarian or demagogue.

nisu na prodaju onom vladaru ili demagogu koji naviše plati.

(Applause)

(Aplauz)

So to go back to that Hollywood paraphrase,

Da se vratim onoj holivudskoj parafrazi,

we do want the prodigious potential

mi želimo da izuzetan potencijal

of artificial intelligence and digital technology to blossom,

veštačke inteligencije i digitalne tehnologije procveta,

but for that, we must face this prodigious menace,

ali za to moramo da se suočimo sa izuzetnom opasnošću,

open-eyed and now.

otvorenih očiju i sada.

Thank you.

Hvala vam.

(Applause)

(Aplauz)