Mark Twain summed up what I take to be one of the fundamental problems of cognitive science with a single witticism. He said, "There's something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment in fact." (Laughter)
Mark Tven je sumirao ono što smatram jednim od temeljnih problema kognitivne nauke samo jednom dosetkom. Rekao je: "Postoji nešto fascinantno u vezi sa naukom. Dobija se veliki obrt pretpostavki od tako sitnog ulaganja u činjenice." (Smeh)
Twain meant it as a joke, of course, but he's right: There's something fascinating about science. From a few bones, we infer the existence of dinosuars. From spectral lines, the composition of nebulae. From fruit flies, the mechanisms of heredity, and from reconstructed images of blood flowing through the brain, or in my case, from the behavior of very young children, we try to say something about the fundamental mechanisms of human cognition. In particular, in my lab in the Department of Brain and Cognitive Sciences at MIT, I have spent the past decade trying to understand the mystery of how children learn so much from so little so quickly. Because, it turns out that the fascinating thing about science is also a fascinating thing about children, which, to put a gentler spin on Mark Twain, is precisely their ability to draw rich, abstract inferences rapidly and accurately from sparse, noisy data. I'm going to give you just two examples today. One is about a problem of generalization, and the other is about a problem of causal reasoning. And although I'm going to talk about work in my lab, this work is inspired by and indebted to a field. I'm grateful to mentors, colleagues, and collaborators around the world.
Tven je mislio to kao šalu, naravno, ali u pravu je, postoji nešto fascinantno u vezi sa naukom. Na osnovu nekoliko kostiju, zaključujemo o postojanju dinosaurusa. Na osnovu spektralnih linija, o sastavu nebula. Od voćne mušice, o mehanizmima nasleđivanja, a na osnovu rekonstruisanih snimaka protoka krvi kroz mozak, ili u mom slučaju, na osnovu ponašanja veoma male dece, pokušavamo da kažemo nešto o osnovnim mehanizmima ljudske kognicije. Konkretno, u mojoj laboratoriji na Odeljenju za mozak i kognitivne nauke na Masačusetskom tehnološkom institutu, provela sam proteklu deceniju pokušavajući da razumem misteriju kako deca uče tako mnogo iz tako malo tako brzo. Jer, ispostavlja se da je fascinantna stvar u vezi sa naukom takođe i fascinantna stvar u vezi sa decom, a to je, da ublažim verziju Marka Tvena, upravo njihova sposobnost da izvuku bujne, apstraktne zaključke brzo i tačno iz oskudnih, izmešanih podataka. Daću vam dva primera. Jedan je o problemu generalizacije, a drugi je o problemu uzročnog rezonovanja. I mada ću govoriti o radu u mojoj laboratoriji, ovaj rad je imao inspiraciju na terenu i njemu ga dugujem. Zahvalna sam mentorima, kolegama i saradnicima širom sveta.
Let me start with the problem of generalization. Generalizing from small samples of data is the bread and butter of science. We poll a tiny fraction of the electorate and we predict the outcome of national elections. We see how a handful of patients responds to treatment in a clinical trial, and we bring drugs to a national market. But this only works if our sample is randomly drawn from the population. If our sample is cherry-picked in some way -- say, we poll only urban voters, or say, in our clinical trials for treatments for heart disease, we include only men -- the results may not generalize to the broader population.
Počeću problemom generalizacije. Uopštavanje na osnovu malih uzoraka podataka je osnovni izvor nauke. Izbrojimo mali deo izbornog tela i predviđamo ishod nacionalnih izbora. Vidimo kako nekolicina pacijenata reaguje na tretman u kliničkom ispitivanju, i donosimo lekove na domaće tržište. Ali ovo funkcioniše samo ako je naš uzorak nasumično izvučen iz populacije. Ako je naš uzorak biran na neki način - recimo, ispitamo samo gradske birače, ili recimo, u kliničkim ispitivanjima tretmana bolesti srca uključimo samo muškarce - rezultati se možda neće generalizovati na širu populaciju.
So scientists care whether evidence is randomly sampled or not, but what does that have to do with babies? Well, babies have to generalize from small samples of data all the time. They see a few rubber ducks and learn that they float, or a few balls and learn that they bounce. And they develop expectations about ducks and balls that they're going to extend to rubber ducks and balls for the rest of their lives. And the kinds of generalizations babies have to make about ducks and balls they have to make about almost everything: shoes and ships and sealing wax and cabbages and kings.
Dakle, naučnike zanima da li su dokazi slučajno uzorkovani ili ne, ali kakve to ima veze sa bebama? Pa, bebe moraju stalno da generalizuju na osnovu malih uzoraka podataka. Vide nekoliko gumenih pataka i nauče da one plutaju, ili nekoliko lopti i nauče da one odskaču. I razvijaju očekivanja u vezi sa patkama i loptama koje će proširiti na gumene patke i lopte do kraja njihovih života. A vrste generalizacija koje bebe prave o patkama i loptama moraju da prave o gotovo svemu: cipelama, brodovima, vosku za pečaćenje, kupusu i kraljevima.
So do babies care whether the tiny bit of evidence they see is plausibly representative of a larger population? Let's find out. I'm going to show you two movies, one from each of two conditions of an experiment, and because you're going to see just two movies, you're going to see just two babies, and any two babies differ from each other in innumerable ways. But these babies, of course, here stand in for groups of babies, and the differences you're going to see represent average group differences in babies' behavior across conditions. In each movie, you're going to see a baby doing maybe just exactly what you might expect a baby to do, and we can hardly make babies more magical than they already are. But to my mind the magical thing, and what I want you to pay attention to, is the contrast between these two conditions, because the only thing that differs between these two movies is the statistical evidence the babies are going to observe. We're going to show babies a box of blue and yellow balls, and my then-graduate student, now colleague at Stanford, Hyowon Gweon, is going to pull three blue balls in a row out of this box, and when she pulls those balls out, she's going to squeeze them, and the balls are going to squeak. And if you're a baby, that's like a TED Talk. It doesn't get better than that. (Laughter) But the important point is it's really easy to pull three blue balls in a row out of a box of mostly blue balls. You could do that with your eyes closed. It's plausibly a random sample from this population. And if you can reach into a box at random and pull out things that squeak, then maybe everything in the box squeaks. So maybe babies should expect those yellow balls to squeak as well. Now, those yellow balls have funny sticks on the end, so babies could do other things with them if they wanted to. They could pound them or whack them. But let's see what the baby does.
Da li bebe zanima da li delić dokaza koji one vide verodostojno predstavlja veću populaciju? Hajde da to otkrijemo. Pokazaću vam dva filma, jedan iz svake od situacija u eksperimentu, i pošto ćete videti samo dva filma, videćete samo dve bebe, a bilo koje dve bebe se razlikuju međusobno na bezbroj načina. Ali ove bebe, naravno, ovde zastupaju grupe beba, i razlike koje ćete videti predstavljaju prosečne grupne razlike u ponašanju beba kroz različite uslove. U svakom filmu ćete videti kako beba radi možda baš ono što biste očekivali da će beba uraditi, a teško da možemo da učinimo bebe čarobnijim nego što već jesu. Ali za mene je čarobna stvar, i ono na šta želim da obratite pažnju, kontrast između ova dva uslova, jer jedino što razlikuje ova dva filma je statistički dokaz koji će bebe primetiti. Pokazaćemo bebama kutiju plavih i žutih lopti, a moja tadašnja studentkinja, sada koleginica na Stenfordu, Jouon Gvon, izvući će tri plave lopte zaredom iz ove kutije, i kada izvuče te lopte, stisnuće ih, a lopte će zapištati. Ako ste beba, to je kao TED govor. Ne može biti bolje od toga. (Smeh) Ali bitna poenta je da je veoma lako izvući tri plave loptice zaredom iz kutije sa pretežno plavim lopticama. Možete to da uradite sa zatvorenim očima. To je verovatno slučajni uzorak iz ove populacije. A ako možete posegnuti u kutiju nasumice i izvaditi stvari koje pište, onda možda sve u toj kutiji pišti. Možda bebe očekuju da žute lopte takođe pište. Te žute lopte imaju zabavne štapiće na kraju, tako da bebe mogu da rade druge stvari sa njima ako hoće. Mogu da ih lupaju ili udaraju. Ali hajde da vidimo šta beba radi.
(Video) Hyowon Gweon: See this? (Ball squeaks) Did you see that? (Ball squeaks) Cool. See this one? (Ball squeaks) Wow.
(Video) Jouon Gvon: Vidiš ovo? (Lopta pišti) Jesi li videla to? (Lopta pišti) Kul. Vidiš ovu? (Lopta pišti) Opa!
Laura Schulz: Told you. (Laughs)
Lora Šulc: Rekla sam vam. (Smeh)
(Video) HG: See this one? (Ball squeaks) Hey Clara, this one's for you. You can go ahead and play. (Laughter)
(Video) JG: Vidiš ovu? (Lopta pišti) Hej Klara, ova je za tebe. Možeš da se igraš. (Smeh)
LS: I don't even have to talk, right? All right, it's nice that babies will generalize properties of blue balls to yellow balls, and it's impressive that babies can learn from imitating us, but we've known those things about babies for a very long time. The really interesting question is what happens when we show babies exactly the same thing, and we can ensure it's exactly the same because we have a secret compartment and we actually pull the balls from there, but this time, all we change is the apparent population from which that evidence was drawn. This time, we're going to show babies three blue balls pulled out of a box of mostly yellow balls, and guess what? You [probably won't] randomly draw three blue balls in a row out of a box of mostly yellow balls. That is not plausibly randomly sampled evidence. That evidence suggests that maybe Hyowon was deliberately sampling the blue balls. Maybe there's something special about the blue balls. Maybe only the blue balls squeak. Let's see what the baby does.
LŠ: Ne moram ni da pričam, zar ne? U redu, lepo je to što će bebe generalizovati osobine plavih loptica na žute loptice, i impresivno je to što bebe mogu da uče imitirajući nas, ali to sve znamo o bebama još odavno. Zaista zanimljivo pitanje je šta se dešava kada pokažemo bebama isto to, a možemo da obezbedimo da bude baš isto jer imamo tajnu pregradu i izvlačimo lopte odatle, ali ovog puta menjamo samo vidljivu populaciju iz koje se izvlači dokaz. Ovoga puta ćemo bebama pokazati tri plave loptice izvučene iz kutije sa pretežno žutim lopticama, i pogodite šta? Verovatno nećete nasumično izvući tri loptice zaredom iz kutije sa većinom žutim lopticama. To nije verovatan slučajno uzorkovani dokaz. Taj dokazi ukazuje da je možda Jouon namerno uzorkovala plave loptice. Možda postoji nešto posebno u vezi sa plavim lopticama. Možda samo plave loptice pište. Hajde da vidimo šta beba radi.
(Video) HG: See this? (Ball squeaks) See this toy? (Ball squeaks) Oh, that was cool. See? (Ball squeaks) Now this one's for you to play. You can go ahead and play.
(Video) JG: Vidiš ovo? (Lopta pišti) Vidiš ovu igračku? (Lopta pišti) O, to je bilo kul. Vidiš? (Lopta pišti) Ova je za tebe da se igraš. Možeš da se igraš.
(Fussing) (Laughter)
(Beba negoduje) (Smeh)
LS: So you just saw two 15-month-old babies do entirely different things based only on the probability of the sample they observed. Let me show you the experimental results. On the vertical axis, you'll see the percentage of babies who squeezed the ball in each condition, and as you'll see, babies are much more likely to generalize the evidence when it's plausibly representative of the population than when the evidence is clearly cherry-picked. And this leads to a fun prediction: Suppose you pulled just one blue ball out of the mostly yellow box. You [probably won't] pull three blue balls in a row at random out of a yellow box, but you could randomly sample just one blue ball. That's not an improbable sample. And if you could reach into a box at random and pull out something that squeaks, maybe everything in the box squeaks. So even though babies are going to see much less evidence for squeaking, and have many fewer actions to imitate in this one ball condition than in the condition you just saw, we predicted that babies themselves would squeeze more, and that's exactly what we found. So 15-month-old babies, in this respect, like scientists, care whether evidence is randomly sampled or not, and they use this to develop expectations about the world: what squeaks and what doesn't, what to explore and what to ignore.
LŠ: Upravo ste videli dve bebe stare 15 meseci koje rade potpuno različite stvari samo na osnovu verovatnoće uzorka koji su zapazile. Dozvolite da vam pokažem eksperimentalne rezultate. Na vertikalnoj osi ćete videti procenat beba koje su stiskale loptu u svakoj situaciji, i kao što ćete videti, mnogo je verovatnije da će bebe generalizovati dokaz kada verodostojnije predstavlja populaciju nego kada je očigledno probran. A to navodi na zabavno predviđanje: recimo da ste izvukli samo jednu plavu loptu iz uglavnom žute kutije. Verovatno nećete izvući tri plave lopte zaredom iz žute kutije, ali biste mogli nasumice uzeti samo jednu plavu loptu. To nije neverovatan uzorak. A ako posegnete u kutiju nasumice i izvučete nešto što pišti, možda sve u kutiji pišti. Dakle, iako će bebe videti mnogo manje dokaza za pištanje, i imati mnogo manje radnji za oponašanje u situaciji sa jednom loptom nego u situaciji koju ste upravo videli, predvideli smo da će bebe stiskati više, i to je upravo ono što smo pronašli. Dakle, bebama od 15 meseci, u ovom pogledu, kao i naučnicima, je bitno da li je dokaz nasumično uzorkovan ili ne, i one koriste to da stvore očekivanja o svetu: šta pišti, a šta ne, šta istražiti, a šta ignorisati.
Let me show you another example now, this time about a problem of causal reasoning. And it starts with a problem of confounded evidence that all of us have, which is that we are part of the world. And this might not seem like a problem to you, but like most problems, it's only a problem when things go wrong. Take this baby, for instance. Things are going wrong for him. He would like to make this toy go, and he can't. I'll show you a few-second clip. And there's two possibilities, broadly: Maybe he's doing something wrong, or maybe there's something wrong with the toy. So in this next experiment, we're going to give babies just a tiny bit of statistical data supporting one hypothesis over the other, and we're going to see if babies can use that to make different decisions about what to do.
Dozvolite mi da vam sada pokažem još jedan primer, ovog puta o problemu uzročnog rasuđivanja. Počinje problemom zbunjujućeg dokaza, koji postoji kod svih nas, a to je da smo deo sveta. I to vam možda ne deluje kao problem ali, kao i većina problema, postaje problem tek kada stvari krenu naopako. Uzmite ovu bebu, na primer. Stvari mu ne polaze za rukom. Želeo bi da pokrene ovu igračku, ali ne može. Pokazaću vam snimak od nekoliko sekundi. Postoje dve mogućnosti, uglavnom. Možda radi nešto pogrešno, ili možda nešto nije u redu sa igračkom. Dakle, u sledećem eksperimentu, daćemo bebama samo delić statističkih podataka koji podržavaju jednu od hipoteza, i videćemo da li bebe mogu to da koriste kako bi donosile različite odluke o onome što će činiti.
Here's the setup. Hyowon is going to try to make the toy go and succeed. I am then going to try twice and fail both times, and then Hyowon is going to try again and succeed, and this roughly sums up my relationship to my graduate students in technology across the board. But the important point here is it provides a little bit of evidence that the problem isn't with the toy, it's with the person. Some people can make this toy go, and some can't. Now, when the baby gets the toy, he's going to have a choice. His mom is right there, so he can go ahead and hand off the toy and change the person, but there's also going to be another toy at the end of that cloth, and he can pull the cloth towards him and change the toy. So let's see what the baby does.
Evo postavke. Jouon će pokušati da pokrene igračku i uspeti u tome. Ja ću potom pokušati dva puta i oba puta neću uspeti, zatim će Jouon pokušati ponovo i uspeti, i to otprilike rezimira odnos koji imam sa mojim studentima po pitanju svih vrsta tehnologija. Ali, ono što je ovde važno jeste to da se pruža malo dokaza da problem nije sa igračkom, već sa osobom. Neki ljudi mogu da pokrenu ovu igračku, a neki ne mogu. Sad, kada beba dobije igračku, imaće izbor. Njegova mama je tu pored, tako da može da joj priđe, preda igračku i promeni osobu, ali na kraju te krpe će biti još jedna igračka, i on može da povuče krpu ka sebi i promeni igračku. Hajde da vidimo šta će beba uraditi.
(Video) HG: Two, three. Go! (Music) LS: One, two, three, go! Arthur, I'm going to try again. One, two, three, go! YG: Arthur, let me try again, okay? One, two, three, go! (Music) Look at that. Remember these toys? See these toys? Yeah, I'm going to put this one over here, and I'm going to give this one to you. You can go ahead and play. LS: Okay, Laura, but of course, babies love their mommies. Of course babies give toys to their mommies when they can't make them work. So again, the really important question is what happens when we change the statistical data ever so slightly. This time, babies are going to see the toy work and fail in exactly the same order, but we're changing the distribution of evidence. This time, Hyowon is going to succeed once and fail once, and so am I. And this suggests it doesn't matter who tries this toy, the toy is broken. It doesn't work all the time. Again, the baby's going to have a choice. Her mom is right next to her, so she can change the person, and there's going to be another toy at the end of the cloth. Let's watch what she does.
(Video) JG: Dva, tri. Sad! (Muzika) LS: Jedan, dva, tri, sad! Arture, pokušaću ponovo. Jedan, dva, tri, sad! JG: Arture, dopusti da ja pokušam ponovo, okej? Jedan, dva, tri, sad! (Muzika) Pogledaj. Sećaš li se tih igračaka? Vidiš te igračke? Da, staviću ovu ovde, a ovu ću ti dati. Možeš da se igraš. LŠ: Okej, Lora, ali naravno, bebe vole svoje mame. Naravno da bebe daju igračke svojim mamama kada ne mogu da učine da prorade. Još jednom, zaista bitno pitanje je šta se dešava kada promenimo statističke podatke neznatno. Ovog puta, bebe će videti kako igračka radi i ne radi potpuno istim redosledom, ali ćemo izmeniti raspodelu dokaza. Ovog puta će Jouon uspeti jednom i neće uspeti jednom, a isto tako ću i ja. Ovo ukazuje da nije bitno ko isprobava igračku, igračka je pokvarena. Ne radi uvek. Još jednom, beba će imati izbor. Njena mama je tu pored, tako da može da promeni osobu, i biće tu još jedna igračka na kraju krpe. Hajde da vidimo šta će uraditi.
(Video) HG: Two, three, go! (Music) Let me try one more time. One, two, three, go! Hmm.
(Video) JG: Dva, tri, sad! (Muzika) Daj da probam još jednom. Jedan, dva, tri, sad! Hmmm.
LS: Let me try, Clara. One, two, three, go! Hmm, let me try again. One, two, three, go! (Music) HG: I'm going to put this one over here, and I'm going to give this one to you. You can go ahead and play. (Applause)
LŠ: Daj da ja probam, Klara. Jedan, dva, tri, sad! Hmmm, daj da probam još jednom. Jedan, dva, tri, sad! (Muzika) JG: Staviću ovu ovde, a ovu ću ti dati. Možeš da se igraš. (Aplauz)
LS: Let me show you the experimental results. On the vertical axis, you'll see the distribution of children's choices in each condition, and you'll see that the distribution of the choices children make depends on the evidence they observe. So in the second year of life, babies can use a tiny bit of statistical data to decide between two fundamentally different strategies for acting in the world: asking for help and exploring. I've just shown you two laboratory experiments out of literally hundreds in the field that make similar points, because the really critical point is that children's ability to make rich inferences from sparse data underlies all the species-specific cultural learning that we do. Children learn about new tools from just a few examples. They learn new causal relationships from just a few examples. They even learn new words, in this case in American Sign Language.
LŠ: Dozvolite da vam pokažem rezultate eksperimenta. Na vertikalnoj osi ćete videti raspodelu izbora dece u svakoj od situacija, i videćete da raspodela izbora koji deca donose zavisi od dokaza koje posmatraju. U drugoj godini života bebe mogu da koriste malo statističkih podataka da bi odabrali između dve fundamentalno različite strategije za postupanje u svetu: pitati za pomoć i istraživati. Upravo sam vam pokazala dva laboratorijska eksperimenta od bukvalno stotina u ovoj oblasti koji imaju sličnu poentu, jer je presudna poenta da se sposobnost dece da donose bogate zaključke iz oskudnih podataka nalazi u osnovi svakog specifičnog kulturnog učenja. Deca uče o novim alatkama na osnovu samo nekoliko primera. Uče nove uzročno-posledične veze iz samo nekoliko primera. Čak uče i nove reči, u ovom slučaju američki znakovni jezik.
I want to close with just two points. If you've been following my world, the field of brain and cognitive sciences, for the past few years, three big ideas will have come to your attention. The first is that this is the era of the brain. And indeed, there have been staggering discoveries in neuroscience: localizing functionally specialized regions of cortex, turning mouse brains transparent, activating neurons with light. A second big idea is that this is the era of big data and machine learning, and machine learning promises to revolutionize our understanding of everything from social networks to epidemiology. And maybe, as it tackles problems of scene understanding and natural language processing, to tell us something about human cognition. And the final big idea you'll have heard is that maybe it's a good idea we're going to know so much about brains and have so much access to big data, because left to our own devices, humans are fallible, we take shortcuts, we err, we make mistakes, we're biased, and in innumerable ways, we get the world wrong. I think these are all important stories, and they have a lot to tell us about what it means to be human, but I want you to note that today I told you a very different story. It's a story about minds and not brains, and in particular, it's a story about the kinds of computations that uniquely human minds can perform, which involve rich, structured knowledge and the ability to learn from small amounts of data, the evidence of just a few examples. And fundamentally, it's a story about how starting as very small children and continuing out all the way to the greatest accomplishments of our culture, we get the world right.
Želim da završim sa samo dve poente. Ako ste pratili moj svet, oblast mozga i kognitivne nauke, poslednjih nekoliko godina, tri ideje su vam privukle pažnju. Prva je da je ovo era mozga. I zaista, bilo je neverovatnih otkrića u neuronaukama: lokalizacija funkcionalno specijalizovanih regija korteksa, dovođenje mišjeg mozga u transparentno stanje, aktiviranje neurona svetlošću. Druga velika ideja je da je ovo era velikih podataka i mašinskog učenja, a mašinsko učenje obećava revoluciju u našem razumevanju svega, od društvenih mreža do epidemiologije. I možda će nam, kako se bavi problemima razumevanja scene i obrade prirodnog jezika, reći nešto o ljudskoj kogniciji. A poslednja velika ideja koju ćete čuti je da je možda dobra ideja da ćemo toliko znati o mozgu i imati toliko pristupa velikim podacima, jer prepušteni sami sebi, ljudi su skloni greškama, koristimo prečice, grešimo, pravimo pogreške, imamo predrasude, i na bezbroj načina, shvatamo svet pogrešno. Mislim da su ovo sve važne priče, i imaju mnogo toga da nam kažu o tome šta znači biti čovek, ali želim da primite k znanju da sam vam danas ispričala veoma drugačiju priču. To je priča o umu, a ne o mozgu, a naročito, to je priča o vrstama proračuna koje jedino ljudski um može da vrši, što podrazumeva bogato, strukturirano znanje i sposobnost učenja iz malih količina podataka, dokaz samo na osnovu nekoliko primera. I u osnovi, to je priča o tome kako počevši kao veoma mala deca i nastavljajući sve do najvećih dostignuća naše kulture, shvatamo svet na pravi način.
Folks, human minds do not only learn from small amounts of data. Human minds think of altogether new ideas. Human minds generate research and discovery, and human minds generate art and literature and poetry and theater, and human minds take care of other humans: our old, our young, our sick. We even heal them. In the years to come, we're going to see technological innovations beyond anything I can even envision, but we are very unlikely to see anything even approximating the computational power of a human child in my lifetime or in yours. If we invest in these most powerful learners and their development, in babies and children and mothers and fathers and caregivers and teachers the ways we invest in our other most powerful and elegant forms of technology, engineering and design, we will not just be dreaming of a better future, we will be planning for one.
Narode, ljudski um ne uči samo iz malih količina podataka. Ljudski umovi smišljaju potpuno nove ideje. Ljudski umovi rađaju istraživanja i otkrića, rađaju umetnost i književnost, poeziju i pozorište, i ljudski umovi se brinu o drugim ljudima: našim starima, mladima, bolesnima. Čak ih i lečimo. U godinama koje su pred nama, videćemo tehnološke inovacije kakve ne mogu ni da zamislim, ali je veoma malo verovatno da ćemo videti bilo šta čak ni približno moći proračuna ljudskog deteta tokom mog života ili vašeg. Ako ulažemo u te najmoćnije učenike i njihov razvoj, u bebe i decu i majke i očeve i staratelje i učitelje onako kako ulažemo u druge naše najmoćnije i najelegantnije oblike tehnologije, inženjeringa i dizajna, nećemo samo sanjati o boljoj budućnosti, već ćemo je planirati.
Thank you very much.
Mnogo vam hvala.
(Applause)
(Aplauz)
Chris Anderson: Laura, thank you. I do actually have a question for you. First of all, the research is insane. I mean, who would design an experiment like that? (Laughter) I've seen that a couple of times, and I still don't honestly believe that that can truly be happening, but other people have done similar experiments; it checks out. The babies really are that genius.
Kris Anderson: Lora, hvala. Ja zapravo imam jedno pitanje za tebe. Pre svega, istraživanje je suludo. Mislim, ko bi osmislio takav eksperiment? (Smeh) Video sam to par puta, i još uvek iskreno ne verujem da se to stvarno dešava, ali i drugi su uradili slične eksperimente; provereno je. Bebe su stvarno toliko genijalne.
LS: You know, they look really impressive in our experiments, but think about what they look like in real life, right? It starts out as a baby. Eighteen months later, it's talking to you, and babies' first words aren't just things like balls and ducks, they're things like "all gone," which refer to disappearance, or "uh-oh," which refer to unintentional actions. It has to be that powerful. It has to be much more powerful than anything I showed you. They're figuring out the entire world. A four-year-old can talk to you about almost anything. (Applause)
LŠ: Znaš, izgledaju stvarno impresivno u našim eksperimentima, ali pomisli na to kako izgledaju u stvarnom životu. Počinje kao beba. Osamnaest meseci kasnije priča sa vama, a bebine prve reči nisu samo one poput lopte i patke, to su i "nema", što se odnosi na nestajanje, ili "o-o", što se odnosi na nenamerne postupke. To mora da je toliko moćno. To mora da je mnogo moćnije od svega što sam vam pokazala. Oni otkrivaju ceo svet. Dete od četiri godine može da priča sa vama o gotovo svemu. (Aplauz)
CA: And if I understand you right, the other key point you're making is, we've been through these years where there's all this talk of how quirky and buggy our minds are, that behavioral economics and the whole theories behind that that we're not rational agents. You're really saying that the bigger story is how extraordinary, and there really is genius there that is underappreciated.
KA: I ako sam te dobro razumeo, druga tvoja ključna poenta je, protekle su godine sa tom pričom o tome kako je um uvrnut i blesav, bihejvioralna ekonomija i čitave teorije o tome kako nismo razumni izvršioci. Ti u stvari govoriš da je veća priča kako je izvanredan, i da je tu zapravo genije koji se potcenjuje.
LS: One of my favorite quotes in psychology comes from the social psychologist Solomon Asch, and he said the fundamental task of psychology is to remove the veil of self-evidence from things. There are orders of magnitude more decisions you make every day that get the world right. You know about objects and their properties. You know them when they're occluded. You know them in the dark. You can walk through rooms. You can figure out what other people are thinking. You can talk to them. You can navigate space. You know about numbers. You know causal relationships. You know about moral reasoning. You do this effortlessly, so we don't see it, but that is how we get the world right, and it's a remarkable and very difficult-to-understand accomplishment.
LŠ: Jedan od mojih omiljenih citata u psihologiji potiče od socijalnog psihologa Solomona Eša, a on je rekao da je osnovni zadatak psihologije da ukloni zavesu samodokazivanja. Postoji milion redova veličine više odluka koje donosite svakodnevno koje pravilno shvataju svet. Imate znanje o predmetima i njihovim osobinama. Prepoznajete ih kada su zaklonjeni. Prepoznajete ih u mraku. Možete da se krećete kroz prostorije. Možete da shvatite šta drugi ljudi misle. Možete da razgovarate sa njima. Možete se kretati u prostoru. Razumete brojeve. Razumete uzročno-posledične veze. Razumete moralno rasuđivanje. Radite to bez napora, tako da se ne vidi, ali to je način na koji poimamo svet, a to je neverovatno dostignuće i veoma teško za razumevanje.
CA: I suspect there are people in the audience who have this view of accelerating technological power who might dispute your statement that never in our lifetimes will a computer do what a three-year-old child can do, but what's clear is that in any scenario, our machines have so much to learn from our toddlers. LS: I think so. You'll have some machine learning folks up here. I mean, you should never bet against babies or chimpanzees or technology as a matter of practice, but it's not just a difference in quantity, it's a difference in kind. We have incredibly powerful computers, and they do do amazingly sophisticated things, often with very big amounts of data. Human minds do, I think, something quite different, and I think it's the structured, hierarchical nature of human knowledge that remains a real challenge.
KA: Pretpostavljam da postoje ljudi u publici koji imaju gledište o ubrzanoj tehnološkoj moći koji bi mogli da ospore tvoju izjavu da nikada za vreme našeg života računar neće uraditi ono što može trogodišnje dete, ali ono što je jasno jeste da u bilo kom scenariju, naše mašine mogu mnogo toga da nauče od naših beba. LŠ: Mislim da je tako. Tu su neki ljudi koji se bave mašinama koje uče. Mislim, nikada se ne treba kladiti protiv beba ili šimpanzi ili tehnologije tek tako, ali nije u pitanju samo razlika u količini, već razlika u vrsti. Imamo neverovatno moćne kompjutere, i oni stvarno obavljaju neverovatno sofisticirane stvari, često sa veoma velikom količinom podataka. Ljudski um čini, po meni, nešto sasvim drugačije, a mislim da je strukturirana, hijerarhijska priroda ljudskog znanja ono što ostaje pravi izazov.
CA: Laura Schulz, wonderful food for thought. Thank you so much.
KA: Lora Šulc, sjajna hrana za misli. Mnogo ti hvala.
LS: Thank you. (Applause)
LŠ: Hvala. (Aplauz)