Sam Harris: Can we build AI without losing control over it?
Sam Harris: Kan vi utveckla artificiell intelligens utan att tappa kontrollen över den?
Sam Harris's work focuses on how our growing understanding of ourselves and the world is changing our sense of how we should live. Full bio
Double-click the English transcript below to play the video.
about a failure of intuition
to detect a certain kind of danger.
upptäcka en viss typ av fara.
most of you will feel
kommer många tycka
is kind of cool.
how the gains we make
to see how they won't destroy us
att se att de inte gör det
to think about these things.
spännande att tänka på.
skulle övertyga dig om
to suffer a global famine,
global svältkatastrof,
or some other catastrophe,
eller någon annan katastrof,
or their grandchildren,
eller deras barnbarn,
on the other hand, is fun,
å andra sidan, är roligt,
about the development of AI at this point
oroar mig mest i utvecklingen av AI
an appropriate emotional response
ett passande emotionellt gensvar
and I'm giving this talk.
och jag håller det här talet.
in building intelligent machines.
av intelligenta maskiner.
just stops getting better for some reason.
hård- och mjukvara inte bättre.
to consider why this might happen.
det skulle kunna hända.
intelligence and automation are,
intelligens och automation,
if we are at all able to.
tekniken, om vi kan.
president of the United States?
destroy civilization as we know it.
civilisationen så som vi känner den.
how bad it would have to be
hur illa det måste vara
improvements in our technology
framsteg inom tekniken
this is the worst thing
behind door number two,
to improve our intelligent machines
våra intelligenta maskiner
machines that are smarter than we are,
maskiner som är smartare än vi,
that are smarter than we are,
the mathematician IJ Good called
det matematikern IJ Good kallade,
as I have here,
armé av ondsinta robotar
will become spontaneously malevolent.
att bli spontant illvilliga.
that we will build machines
att bygga maskiner
more competent than we are
between their goals and our own
mellan deras mål och våra
we take pains not to harm them.
för att inte skada dem.
a building like this one,
one day build machines
kommer att bygga maskiner
med samma förakt.
far-fetched to many of you.
långsökt för många av er.
that superintelligent AI is possible,
på att superintelligent AI är möjlig,
with one of the following assumptions.
några av följande antaganden.
processing in physical systems.
av information i fysiska system.
than an assumption.
än bara ett antagande.
narrow intelligence into our machines,
i många maskiner,
intelligence already.
mänsklig intelligens.
"general intelligence,"
kallas "generell intelligens"
across multiple domains,
över flera dimensioner,
har klarat det. Eller hur?
to build systems of atoms
bygga system av atomer
intelligent behavior,
unless we are interrupted,
om vi inte blir avbrutna,
build general intelligence
that the rate of progress doesn't matter,
utvecklingstakten inte spelar någon roll,
is enough to get us into the end zone.
räcker för att nå slutstationen.
We don't need exponential progress.
behöver inte vara exponentiell.
is that we will keep going.
att vi kommer fortsätta.
our intelligent machines.
av de intelligenta maskinerna.
the source of everything we value
summan av allt vi sätter värde på
everything we value.
skydda det vi värdesätter.
that we desperately need to solve.
like Alzheimer's and cancer.
Alzheimers och cancer.
We want to improve our climate science.
Vi vill utveckla klimatforskningen.
and there's no brake to pull.
och det finns ingen broms.
on a peak of intelligence,
nått intelligensens topp,
our situation so precarious,
vår situation så vansklig,
about risk so unreliable.
riskantaganden så opålitliga.
who has ever lived.
som någonsin levt.
is John von Neumann.
är John von Neumann.
made on the people around him,
gjorde på folk runt honom,
mathematicians and physicists of his time,
och fysikerna under hans levnad,
about him are half true,
om honom är halvsanna,
who has ever lived.
more depressing than it needs to be.
mer deprimerande än det redan är.
that the spectrum of intelligence
att spektrat av intelligens,
than we currently conceive,
än vi kan föreställa oss idag,
that are more intelligent than we are,
som är intelligentare än vi,
explore this spectrum
utforska detta spektrum
that we can't imagine.
sätt som vi inte kan förstå.
this is true by virtue of speed alone.
genom hastigheten det sker med.
a superintelligent AI
byggt en superintelligent AI
than your average team of researchers
genomsnittliga forskargrupp
function about a million times faster
en miljon gånger snabbare
about a million times faster
en miljon gånger snabbare
of human-level intellectual work,
20 000 år av mänskligt tänkande,
much less constrain,
of superintelligent AI
en superintelligent AI
the first time around.
the perfect labor-saving device.
den perfekta "jobbspararen".
that can build the machine
som kan bygga maskinen
of raw materials.
kostnaden för råmaterialet.
the end of human drudgery.
av mänsklighetens slit.
of most intellectual work.
av intellektuellt arbete.
do in this circumstance?
under sådana omständigheter?
and give each other massages.
och ge varandra massage.
questionable wardrobe choices,
underliga klädesplagg,
could be like Burning Man.
and political order?
ekonomiska och politiska system?
and unemployment
och tillgång till arbete
to immediately put this new wealth
låta detta nya välstånd
the covers of our business magazines
omslagen på våra affärstidningar
would be free to starve.
or the Chinese do
in Silicon Valley
företag i Silicon Valley
superintelligent AI?
of waging war,
kapabel att föra krig,
of the competition here
of this kind of breakthrough
kring denna typ av genombrott
that AI researchers say
we're told not to worry is time.
vi inte ska oroa oss handlar om tid.
don't you know.
about overpopulation on Mars."
ett överbefolkat Mars."
pretty little head about it."
of information processing,
om att processa information,
utveckla våra maskiner,
some form of superintelligence.
någon form av superintelligens.
how long it will take us
hur lång tid det kommer att ta
to do that safely.
att göra det riskfritt.
to do that safely.
50 years is not what it used to be.
men 50 år är inte vad det brukade vara.
has been on television.
our species will ever face.
mänskligheten någonsin skådat.
to have an appropriate emotional response
ett passande emetionellt gensvar
to believe is coming.
anledning att tro kommer.
has a nice analogy here.
drar en intressant parallell.
a message from an alien civilization,
meddelande från en främmande civilisation,
på er planet om 50 år.
the months until the mothership lands?
tills moderskeppet landar?
more urgency than we do.
brådskar mer än vad vi gör.
can't help but share our values
att dela våra värderingar
extensions of ourselves.
vara en förlängning av oss.
become their limbic systems.
deras limbiska system.
and only prudent path forward,
förståndigaste vägen framåt,
directly into our brains.
direkt i våra hjärnor.
and only prudent path forward,
och smartaste vägen att ta,
about a technology
säkerhetsriskerna med en teknik
before you stick it inside your head.
man stoppar in den i hjärnan.
building superintelligent AI on its own
utveckla en fristående superintelligent AI
integrate our minds with it.
integrera våra sinnen med den.
and governments doing this work
stater som står för vetenskapen
as being in a race against all others,
vara i kamp med alla andra,
is to win the world,
vinner man världen,
in the next moment,
förstör den i nästa steg,
that whatever is easier to do
allt som är enklare att göra
I don't have a solution to this problem,
på det här problemet,
that more of us think about it.
måste fundera på det.
like a Manhattan Project
we'll inevitably do that,
det tror jag oundvikligen vi gör,
how to avoid an arms race
undvika en kapprustning
that is aligned with our interests.
med våra egna intressen.
about superintelligent AI
to get the initial conditions right,
en chans att få det rätt från början,
consequences of getting them right.
politiska konsekvenserna.
is the source of intelligence,
är källan till intelligens,
is what the basis of intelligence is,
är basen för intelligens,
these systems continuously,
dessa system kontinuerligt,
of cognition very likely far exceeds
kognitionshorisonten vida överstiger
of building some sort of god.
skapa någon form av gud.
ABOUT THE SPEAKER
Sam Harris - Neuroscientist, philosopherSam Harris's work focuses on how our growing understanding of ourselves and the world is changing our sense of how we should live.
Why you should listen
Sam Harris is the author of five New York Times bestsellers. His books include The End of Faith, Letter to a Christian Nation, The Moral Landscape, Free Will, Lying, Waking Up and Islam and the Future of Tolerance (with Maajid Nawaz). The End of Faith won the 2005 PEN Award for Nonfiction. Harris's writing and public lectures cover a wide range of topics -- neuroscience, moral philosophy, religion, spirituality, violence, human reasoning -- but generally focus on how a growing understanding of ourselves and the world is changing our sense of how we should live.
Harris's work has been published in more than 20 languages and has been discussed in the New York Times, Time, Scientific American, Nature, Newsweek, Rolling Stone and many other journals. He has written for the New York Times, the Los Angeles Times, The Economist, The Times (London), the Boston Globe, The Atlantic, The Annals of Neurology and elsewhere. Harris also regularly hosts a popular podcast.
Harris received a degree in philosophy from Stanford University and a Ph.D. in neuroscience from UCLA.
Sam Harris | Speaker | TED.com