Michio Kaku professor of theoretical physics of the City College and the Graduate Center of the City University of New York and this is exploration every week an exploration we discuss the fascinating world of science and its impact on society and today leading off we’re gonna summarize some of the big stories in science first of.
All on our lead story we’re gonna go to the movies that’s right we’re gonna go to the movies where there’s a new movie called Blade Runner 20:49 that’s attracting a lot of media attention so far the box-office draw has not been as great as expected however the reviews were fantastic and the question is.
Well Blade Runner is based on the idea that one day robots will be pretty much like human beings and the question is how close was Blade Runner to the actual truth the first movie came out in the 1980s and it predicted that by the year 2019 two years from now in 2019 robots would be pretty much like us now the new movie has it 30 years later 20:49 and so the.
Question is what went wrong I mean we had movies like 2001 which came out in 1968 and boy was that movie also wrong so why were all these projections made in the 60s 70s 80s why were they all wrong and for that matter when might they be realize when might we have robots that look well.
Just like us and also a cosmic mystery.
Has been solved in outer space just recently scientists observed the collision of two neutron stars in deep space.
And we found bingo that’s the origin of gold now gold of course is pretty much everywhere in your wedding ring jewelry and Fort Knox but the question that has intrigued scientists for generations is where does gold come from I mean during the Middle Ages alchemists tried to create gold out of lead using something called the Philosopher’s Stone well today we don’t believe in that of course.
Because we have some thing called nuclear physics but that left open the mystery where does gold and platinum where do they come from anyway they don’t seem to fit any known sequence of events but now we have the clue from the gravity wave detectors which are now being used as.
Telescopes to peer into the heart of black holes and neutron stars and also we’ll say a few things about cancer therapy and immunotherapy is maturing very fast and could be the next major weapon in our arsenal against cancer cancer is the second leading.
Cause of death in the United States so we’ll say a few things about the new therapies that attack cancer at the genetic and cellular level well let’s just jump right into some of the top.
Stories of the past week as I said before we’re gonna go to the movies a lot of us love the movies however many times movie makers.
Even though they have the movie magic of special effects sometimes they.
Get the science a little bit wrong back in 1968 we had that path-breaking film 2001 it was a real eye-opener but it predicted that back in the 1990s we would already begin to have robots that we can talk to and of.
Course the movie was said in the year 2001 now we.
Expect that will perhaps have the first moon base man moon base perhaps in the next 15 years or so and so the movie was off by perhaps 50 to 70 years and so then the question is why and for that matter Blade Runner which was the pioneering film which came out at.
The 1980s it predicted that by the year 2019 we would have replicants that is robots to do the dirty work of humans in outer space however these robots look just like us they reason like us talk like us in fact they are pretty much indistinguishable from us and if you read the movie reviews there’s even a theory that Harrison.
Ford was really a robot all along and so the question is what went.
Wrong with all these predictions well back in the 1950s when artificial intelligence first got off the ground things look so easy computers are becoming more powerful that could begin to play checkers then they begin to play chess then they begin to play more complicated games and.
As a consequence mathematicians got well hyped up they thought that within.
A summer or two they said we’re gonna begin to have robots that are just as smart as us well boy were they wrong.
So what was the origin of this mistake.
That lasted for 50 years but fifty years ago scientists made a huge mistake in artificial intelligence they assumed that the human brain was a digital computer however if you think about it the human brain actually has very little in common with a digital computer first of all the brain has no programming no programming that we’re aware of.
You’ve taken apart the brain and sure enough we don’t see windows there we don’t say Fortran we don’t see Bell go we don’t see any computer language inside the.
Brain so there’s no programming second of all computers have a chip like a Pentium chip but there’s no Pentium chip in the brain you can pick a parts of brain the run burner on and sure enough there are no central processors no CPUs inside the brain next you can remove half the brain and the brain still functions in fact there was a young girl.
Once who complained about headaches they scanned her brain and they were shocked to find that she had literally half a brain so the brain can function.
With 50% of his mass missing however if you have one transistor out of.
Place inside a computer the whole computer will malfunction so the brain has no subroutines it is no.
Central processor no Pentium chip no programming and then the question is what is it there for well that was a mistake thinking that digital computers would be the guiding path to unlock the secrets of the human brain and artificial intelligence boy were we wrong so the approach back then was.
Approach that is let’s write a computer program with all the laws of mother nature in it so that you plug this disk into a computer and the computer suddenly says I am aware I think I am I am like a human that was the goal a top-down approach that is one gigantic disk with all the lines of intelligence on it you insert it into a computer.
And bingo you have a robot like how in the movie 2001 well wrong now we realize that the brain is not a digital computer at all it’s something totally different it’s a learning machine called a neural network it literally rewires itself after learning every task so now 50 years after we made that mistake back to the 1950s scientists are introducing something.
Called deep learning deep learning is a program that actually.
Learns a little bit slow as it may seem however is something that we should have investigated 50 years ago I mean think about it.
As stupid as it was yesterday your laptop never learns anything it makes the same mistakes over and over and over again but you see our brain operates on the bottom-up approach it constantly makes mistakes.
It bumps into things it learns from these mistakes and moves on we all know the secret to go to Carnegie Hall if you are a musician the secret is practice practice practice you see with every practice your own neural nets become closer and closer to the correct one and that’s how we learn by trial and error by bumping into things so now scientists are slowly beginning to adopt the bottom up.
And the top-down approach because after all that’s how we learn in high school and college the bottom-up approach is to bump into things and learn things the hard way the school of hard knocks the top-down approaches learning from a book.
Like learning calculus from a book and so the combination of the two bottom-up and top-down approach is.
Perhaps the way to go so when might we have robots that are just like us well I think we were way off I think perhaps late in this century we may begin to.
Approach the intelligence of a human being right now our robots are as smart as a cockroach as smart as a bug even bugs can zip around obstacles find mates find food our most advanced robots you put them in the middle of a straight what do.
They do they get lost that’s what they do.
However eventually they’ll be as smart as a.
Mouse then as smart as a rat then as smart as a rabbit then as smart as a cat a dog and.
Finally perhaps by the end of.
The century as smart as a monkey at that point they could be dangerous just like in the movie 2001 we had a murderous.
Robot that took over the functions of a rocket ship yeah that’s possible I don’t think it’s possible too late in the century as Rodney Brooks former director of the MIT famous artificial intelligence laboratory once told me on exploration he said that the probability that someone’s gonna invent a humanoid robot anytime soon is equivalent to the probability of a hurricane creating a Boeing c-47 jet out of debris and of course it ain’t gonna happen anytime soon however.
It could happen in the future in which case humans.
Bit about the job market because of course robots as they get smarter will start to replace human type jobs but some jobs will be out of touch of robots for quite a while for example semi-skilled non-repetitive.
Workers like plumbers like carpenters like construction.
Workers like gardeners like policemen they require a level of pattern recognition and common-sense that’s way beyond anything that we can create in the laboratory so these.
Kinds of jobs specialized jobs for semi skilled workers those jobs are gonna be with us for a long time to come the real losers in this game is a repetitive workers workers do they do the same motion over and over again those jobs are on the.
Way out also jobs that involve search engines like paralegals those jobs are on the way out and also middlemen jobs you got to be careful brokers for example they’re gonna be on the way out unless they add something to their menu of abilities that is intellectual capital that is expertise that is knowledge experience imagination innovation know-how these are the things that robots don’t have.
Brokers are right in the crosshairs of the computer revolution they will survive only if they begin to adopt intellectual capital capital of the mind rather than using your hands and building on commodities and so in some sense robots are going to become more proliferating into the job market but.
That robots cannot do the three things are first of all work with humans and understand human behavior and personality that’s why we’re gonna have lawyers in the future lawyers can talk to juries lawyers can talk to judges they can argue cases they can appeal to the sensibilities of the jury robots cannot robots are clueless when it comes to human relations because well quite frankly human relations are so complicated so that’s one area where robots cannot excel.
Another area is pattern recognition they don’t understand what they see they can barely see the line circle squares inside a room rather than seeing table and stool and book they don’t see objects for what they are pattern recognition is still very weak and common-sense is also very weak they don’t know that water is wet they don’t know that when you die you don’t come back the next day they don’t know the mothers are older than their daughters now everybody knows that yeah humans know.
That but robots do not now I’m not saying that you can’t get a robot as intelligent as a human being I’m just.
Saying that the problem is much more difficult than people thought back in the 1950s when they first created something called.
Artificial intelligence we now realize that the brain is very complicated it’s a parallel processor it does not have a Pentium chip at all it does trillions of calculations simultaneously while a computer does one calculation at a time now you’ve probably seen some calculations showing that by 2045 or so robots are going to be as smart as us and beyond that better.
Than us well not so fast maybe maybe not but that calculation is based on something called Moore’s Law Moore’s law says the computer.
Power doubles every 18 months but Moore’s law that we depend upon to have Christmas present twice as powerful as the previous Christmas that law is slowing down that’s right what computer power is no longer doubling at the rate of 18 months per doubling cycle so eventually perhaps in.
10-15 years Moore’s law could collapse why because computer chips like a Pentium chip has a layer baby 20 atoms across well in five to ten years.
That layer is gonna be five atoms across and at that point the Heisenberg uncertainty principle kicks in you don’t know where the electron is anymore and the whole thing goes kaput so we have to create a new generation of computers optical computers DNA computers quantum.