Nick Bostrom: What happens when our computers get smarter than we are?
Nik Bostrom (Nick Bostrom): Šta će se desiti kad naši kompjuteri postanu pametniji od nas?
Nick Bostrom asks big questions: What should we do, as individuals and as a species, to optimize our long-term prospects? Will humanity’s technological advancements ultimately destroy us? Full bio
Double-click the English transcript below to play the video.
philosophers and computer scientists,
filozofa i informatičara
the future of machine intelligence,
i razmišljamo o budućnosti
things are sort of science fiction-y,
kao naučna fantastika,
human condition.
guests on this planet,
skoro pristigli gosti
was created one year ago,
pre godinu dana.
would be 10 minutes old.
bila stara 10 minuta,
two seconds ago.
pre dve sekunde.
world GDP over the last 10,000 years,
u poslednjih 10.000 godina.
to plot this for you in a graph.
da vam ovo predstavim grafički.
for a normal condition.
to je neobičan oblik.
da sednem na njega.
of this current anomaly?
ove savremene anomalije?
through human history,
tokom ljudske istorije,
advances extremely rapidly --
so very productive.
to the ultimate cause.
do prvobitnog uzroka.
distinguished gentlemen:
tokens, an incredible feat.
što je neverovatan podvig.
superstring revolution.
revoluciju superstruna.
this is what we find:
pronaći ćemo ovo:
in the exact way it's wired.
u načinu na koji je povezana.
be too complicated, however,
previše komplikovane,
been 250,000 generations
take a long time to evolve.
treba mnogo vremena da se razviju.
to intercontinental ballistic missiles.
do interkontinentalnih raketa.
that everything we've achieved,
da sve što smo postigli
changes that made the human mind.
koje su stvorile ljudski um.
is that any further changes
the substrate of thinking
enormous consequences.
ogromne posledice.
think we're on the verge
da se nalazimo nadomak
a profound change in that substrate,
duboku promenu u toj osnovi,
about putting commands in a box.
kutija u koju ste ubacivali komande.
handcraft knowledge items.
for some purposes,
you couldn't scale them.
niste mogli da ih podešavate.
what you put in.
ono što ste već imali ubačeno.
in the field of artificial intelligence.
na polju veštačke inteligencije.
around machine learning.
representations and features,
i njegove komponente,
often from raw perceptual data.
iz sirovih opažajnih podataka.
that the human infant does.
to radi novorođenče.
limited to one domain --
koja nije ograničena na jednu oblast -
between any pairs of languages,
između bilo koja dva jezika
on the Atari console.
bilo koju igricu na Atariju.
the same powerful, cross-domain
da može kao i ljudi
as a human being has.
više oblasti istovremeno.
algorithmic tricks
how to match in machines.
da postignemo kod mašina.
to match those tricks?
da to postignemo?
leading A.I. experts,
vodeće stručnjake za V.I.
and one of the questions we asked was,
i jedno od naših pitanja je bilo:
there is a 50 percent probability
da postoji 50% šanse
human-level machine intelligence?"
inteligenciju koja je na nivou ljudske?"
as the ability to perform
kao sposobnost da se obavi
as an adult human,
bi to uradila odrasla osoba.
within some limited domain.
a ne u nekoj ograničenoj oblasti.
oko 2040. ili 2050,
group of experts we asked.
grupu stručnjaka pitali.
much later, or sooner,
mnogo kasnije, ili ranije.
limit to information processing
konačna granica obrade informacija
the limits in biological tissue.
izvan granica biološkog tkiva.
at 200 hertz, 200 times a second.
na 200 herca, 200 puta u sekundi.
operates at the Gigahertz.
radi na jednom gigahercu.
100 meters per second, tops.
najviše 100 m u sekundi.
at the speed of light.
da putuju brzinom svetlosti.
to fit inside a cranium,
da bi stao u lobanju,
of a warehouse or larger.
veličine skladišta ili veći.
lies dormant in matter,
čeka uspavan,
lay dormant throughout human history,
bila uspavana tokom ljudske istorije,
the power of artificial intelligence.
moć veštačke inteligencije.
an intelligence explosion.
da vidimo eksploziju inteligencije.
about what is smart and what is dumb,
šta je pametno, a šta glupo,
roughly like this.
or whoever your favorite guru is.
ili nekog vašeg omiljenog gurua.
of artificial intelligence,
veštačke inteligencije,
probably more like this:
at zero intelligence,
na nultoj inteligenciji
years of really hard work,
godina vrednog rada,
mouse-level artificial intelligence,
do veštačke inteligencije na nivou miša,
cluttered environments
u zakrčenim prostorima
of really hard work, lots of investment,
vrednog rada i mnogo ulaganja,
chimpanzee-level artificial intelligence.
of really, really hard work,
artificial intelligence.
we are beyond Ed Witten.
pretekli smo Eda Vitena.
at Humanville Station.
kad pretekne ljude,
samo prošišati pored.
to questions of power.
twice as strong as a fit human male.
snažnija od čoveka.
and his pals depends a lot more
i njegovih drugara zavisi više
what the chimpanzees do themselves.
nego od toga šta šimpanze rade.
on what the superintelligence does.
šta superinteligencija radi.
that humanity will ever need to make.
koji će čovečanstvo morati da napravi.
at inventing than we are,
u pronalazaštvu od nas
on digital timescales.
vremenskim rokovima.
a telescoping of the future.
that you could have imagined
koju možete da zamislite
in the fullness of time:
of minds into computers,
ili čuvanje umova u kompjuterima,
with the laws of physics.
develop, and possibly quite rapidly.
da napravi, i to verovatno jako brzo.
technological maturity
tehnološkom zrelošću
it would be able to get what it wants.
mogla bi da dobije šta hoće.
be shaped by the preferences of this A.I.
koja bi bila oblikovana po želji te V.I.
what are those preferences?
avoid anthropomorphizing.
every newspaper article
jer svaki novinski članak
has a picture of this:
to conceive of the issue more abstractly,
problem apstraktnije,
as an optimization process,
kao proces optimizacije,
into a particular set of configurations.
u određeni set konfiguracija.
a really strong optimization process.
snažan proces optimizacije.
available means to achieve a state
dostupnih sredstava da postigne stanje
conenction between
obavezno veza između:
would find worthwhile or meaningful.
smatramo važnim ili smislenim.
to make humans smile.
zadatak da nasmeje ljude.
or amusing actions
korisne ili zabavne radnje
effective way to achieve this goal:
da postigne taj cilj,
muscles of humans
na mišiće ljudskih lica
a difficult mathematical problem.
da reši težak matetematički problem.
to get the solution to this problem
za rešenje ovog problema
into a giant computer,
u džinovski kompjuter,
kapacitet za razmišljanje.
an instrumental reason
instrumentalni razlog
might not approve of.
možda ne slažemo.
problem from being solved.
za rešavanje matematičkog problema.
go wrong in these particular ways;
baš na ovaj način;
optimization process
proces optimizacije
maksimalan učinak za cilj X,
that your definition of x
do čega vam je stalo.
in many a myth.
uči i jedan mit.
he touches be turned into gold.
bude pretvoreno u zlato.
she turns into gold.
i ona se pretvori u zlato.
za ono što bi se desilo
optimization process
moćan proces optimizacije
or poorly specified goals.
sticking electrodes into people's faces,
da kači elektrode ljudima na lica,
if we've grown dependent on the system --
ako bismo bili zavisni od sistema -
to the Internet?
za gašenje interneta?
flicked the off switch to humanity,
prekidač čovečanstvu,
for example, right here.
na primer, ovde.
an intelligent adversary;
inteligentan neprijatelj;
i planirati kako da ih izbegnemo.
and plan around them.
superinteligencije,
at that than we are.
that we have this under control here.
tako sigurni da imamo kontrolu ovde.
a little bit easier by, say,
da malo olakšamo posao
staviti V.I. u kutiju,
from which it cannot escape.
iz koje ona ne može pobeći.
the A.I. couldn't find a bug.
da V.I. neće naći rupu?
find bugs all the time,
stalno nalaze rupe,
to create an air gap,
da bismo stvorili vazdušni zid,
using social engineering.
pomoću socijalnog inženjeringa.
out there somewhere
postoji neki zaposleni
her account details
da mu da podatke o svom računu,
from the I.T. department.
da je iz IT sektora.
ako ste vi V.I.,
around in your internal circuitry
u vašem unutrašnjem sistemu
can use to communicate.
koje možete koristiti za komunikaciju.
da ste pokvareni
you up to see what went wrong with you,
da vide šta nije u redu,
to a really nifty technology,
na pogodan tehnološki uređaj
that the A.I. had planned.
koji je V.I. planirala.
not be confident in our ability
locked up in its bottle forever.
zaključanog u boci zauvek.
is to figure out
such that even if -- when -- it escapes,
čak i ako pobegne
fundamentally on our side
jer je suštinski na našoj strani
this difficult problem.
za ovaj težak problem.
that this problem can be solved.
može da se reši.
a long list of everything we care about,
svega onoga do čega nam je stalo
in some computer language
na nekom kompjuterskom jeziku
that uses its intelligence
koja koristi svoju inteligenciju
in such a way that it is motivated
je konstruisan tako da je motivisan
that it predicts we would approve of.
za koje predviđa da ih mi odobravamo.
its intelligence as much as possible
što je više moguće
very good for humanity.
for the intelligence explosion
in just the right way
na tačno određen način
kontrolisanu detonaciju.
need to match ours,
se poklope sa našim
how the A.I. behaves,
kako se V.I. ponaša,
that the A.I. might encounter
u kojima se V.I. može naći
that would need to be solved, sorted out:
koja bi trebalo rešiti:
uncertainty and so forth.
to be solved to make this work
trebalo rešiti da bi ovo funkcionisalo
a superintelligent A.I.,
superinteligentne V.I.,
is a really hard challenge.
je vrlo težak izazov.
koja je bezbedna
challenge on top of that.
how to crack the first challenge
kako da razbije prvu teškoću,
the additional challenge
work out a solution
by the time it is needed.
kad dođe vreme.
the entire control problem in advance
ceo problem kontrole unapred,
can only be put in place
architecture where it will be implemented.
na kojoj bi bilo primenjeno.
that we solve in advance,
rešimo unapred,
to the machine intelligence era
ka dobu mašinske inteligencije
that is well worth doing
koju bi vredelo uraditi
things turn out okay,
da ako stvari ispadnu okej,
look back at this century
pogledati unazad na ovo doba
the one thing we did that really mattered
ono što je bilo zaista važno
ABOUT THE SPEAKER
Nick Bostrom - PhilosopherNick Bostrom asks big questions: What should we do, as individuals and as a species, to optimize our long-term prospects? Will humanity’s technological advancements ultimately destroy us?
Why you should listen
Philosopher Nick Bostrom envisioned a future full of human enhancement, nanotechnology and machine intelligence long before they became mainstream concerns. From his famous simulation argument -- which identified some striking implications of rejecting the Matrix-like idea that humans are living in a computer simulation -- to his work on existential risk, Bostrom approaches both the inevitable and the speculative using the tools of philosophy, probability theory, and scientific analysis.
Since 2005, Bostrom has led the Future of Humanity Institute, a research group of mathematicians, philosophers and scientists at Oxford University tasked with investigating the big picture for the human condition and its future. He has been referred to as one of the most important thinkers of our age.
Nick was honored as one of Foreign Policy's 2015 Global Thinkers .
His recent book Superintelligence advances the ominous idea that “the first ultraintelligent machine is the last invention that man need ever make.”
Nick Bostrom | Speaker | TED.com