Class: Informatics, Computing, and the Future
Instructor: Dan Berleant
Transcriber: Brooke Yu
Date: Tuesday, April 09, 2013
Professor: Okay folks, so here's where we are in the course. We're getting real close to where the student presentations start. In fact, so many people registered for the course that we'll do three presentations per class. But I don't think we should do that. Let's change them to be shorter so we'll get 4 people per class. They'll be 15 minutes each.
If all four people want to spend 20 minutes, we'll go over time so we won't do that. We'll try to get 4 people per class. We can take people any time. If we do the last four classes, that would work. If you want to go before then, that will work. If you really want to feel you'd be better off doing your presentation during the final- there won't be an exam- but you can come to class and do your presentation, if you like. I know the homework that's due on the 16th, which is a week from today, asks you to pick a slot, but that won't be soon enough.
I'm going to pass around some sheets of paper, so if you haven't signed up, go ahead and pick a day. Pick three from most favored to third favored and I'll see what I can do to get you into your favorite slot.
It's all written on the other side. It's just scrap paper. Use the other side.
Male Student: Our group- we're thinking about doing ours in a group of 4.
Male Student: Do we have to present for a full hour?
Professor: Yeah, 15 minutes per person.
Male Student: Oh, okay.
Professor: Okay, so pick your first choice, second choice, and third. Any time between next time until the final exam period.
Oh, and write your name, of course.
And when you're done, just hold them up and I'll come get them.
Male Student: Can we go this Thursday?
Professor: Yeah, you can go then if you want.
Anybody else want to hand theirs in?
Male Student: I don't think we'll hit an hour, so we'll probably just present individually.
Professor: Any more forms? Alright.
I can't guarantee that if everyone signed up for the same days you'll get the day you want. A few people are not here, and they're not signing up, so they'll get last choice. Any questions about anything at this point?
Alright, so where are we? We're going to finish talking about the singularity today. Supposedly it's not coming for 30 years.
Did anybody sign up for Thursday? We'll do some robotics on Thursday or whatever. I'll have to rearrange based on when you signed up.
Male Student: We just had to write down three dates, right?
Male Student: Alright.
Professor: Okay, so at this point we're moving into the end of the semester. I was about to apologize for cramming you down to fifteen minutes, but I don't think too many people are upset about it. More people are in this class than last time, so we have to schedule accordingly.
Let's talk about the singularity. That's what the movie was about, right? What's your conception of the singularity? Or what did you take away from the movie? What's your person concept of the singularity?
Well, let me ask this- according to his projected timeline, will the singularity be in your lifetime? Yeah. Your life expectancy is such that it will be.
In fact, he's even got it so that assuming he gets to live for a long time- he thinks it'll be in his lifetime. So a lot of people who want to see the profits of the singularity- they all think it's going to happen when they're alive. So that makes you... I think it detracts from the for how much you can believe it.
They kind of change their thoughts accordingly.
But with to in mind, let's see. So anybody remember what the singularity was from the movie?
Male Student: Like the description of it?
Male Student: Eventually we'll enter a time frame when technology have evolved so much that it could replace life and the fact that, you know, we can transcend humanity- like think of... there's an aspect of immortality. Technology will have grown so fast to become that.
Professor: Anyone else want to add details? Yeah, that's sort of what he was talking about. You ask what that all boils down to- technology, as they say, will change our world so much that we really can't predict what it's going to be like. I've shown you models like the exponential curve, but if there's a singularity, you can't know what's going to be after that.
If things go up too fast, you don't know what will happen. That's what the singularity means- things will be so different that you can't predict what they'll be like. But they try, and they come up with ideas like immortality.
Well, it's kind of a fuzzy and intuitive concept. Let's try to break it down.
One kind of singularity is the technological singularity.
It simply says that technology will change faster and faster until it has changed so much so quickly that we can't predict what it will be like. It'll be different. That's pretty vague.
There's another called the AI singularity.
Has anyone ever heard that term?
Did anyone read the story yet that I passed out? Okay, so that's really the AI singularity in that story. So it says that if we can make a computer that's smarter than a person, that means we can create something smarter than ourselves. Well, what could that entity do? If we could create something smarter than us, then it could create something smarter than it. If we could reach that point, the process will just continue- it'll step-wise continue because each computer will build even smarter computers and it will spiral out of control and who knows where it will end. We don't know. It's a little different because it says there will be no known limit on the robots. That's kind of a brainstopper. It's really weird, you know?
I mean, it means if we can make a computer that's 1% smarter than a person, then there will be computers any amount smarter than a person because it'll keep going.
Male Student: Yeah, what about computers that just learn from each other and just exponentially grow smarter from each other?
Professor: Now you're making it more complicated.
Male Student: I mean, if someone made one that could learn from themselves.
Professor: You all get smarter everyday because you're in school and have life experiences. A person's intelligence increases over time. So there are those details that aren't modeled by this simple concept of the spiraling.
So what do you think? Do you think an AI singularity will happen?
If you were to guess, how many would say yes, no, and not sure?
How many would vote for yes?
How many for no?
How many for maybe? 6.
Interesting. Of course, many of you are computer students. So no one voted for no, but most people are not sure. So it might, right? And it'll be weird. Everyone agree with that? Life will be different. So by a vote of 9 to 0, people think life may be very different at some point. I guess 3 people will say life will be different, and 6 say it may be different.
Well, let me give you some notes.
The root of the concept of what is the singularity comes from physics and mathematics. It's intended to capture the idea that you can make a mathematical calculation and there's no answer. Has anyone ever seen this before on their calculator? It stands for not a number
If you try to divide by 0 or something, you might get that.
I don't know what excel does. Let's find out. I'm going to do this. It didn't give me a number, and it didn't say NaN. It just said divided by 0. Because it's not a numerical answer either.
Okay, so let's, you know, in physics they like to model the universe using mathematics, right?
Let's suppose we're interested in a property of matter. Let's call it density.
And density is... what's density?
Male Student: Mass in a given volume.
Professor: Right. So to calculate density, we could divide mass by volume. So if you have 10 pounds in 1 gallon, its density is 10 pounds per gallon, which is probably more than a gallon of milk.
Well, what does milk weigh in a gallon?
Male Student: A little more than 8 pounds.
Professor: Okay. Any kind of oil floats on water, so it would be less than 8 pounds.
Now if you have gas, you can squish the gas, right? Like in your car. When the piston squeezes the air and gas... you know how the piston works? The piston squeezes it and the spark plug gets hot which pushes the piston. So if you squeeze gas, the volume decreases by the mass is the same. If you squish a solid or liquid, it doesn't squish easily, but it can be squished and it gets denser. If the bottom number gets smaller, the over all fraction gets bigger. Density gets up. If you have this in 1/2 a gallon, you have the same amount of mass in a smaller space
If you squish it until it's 0, what's the density?
Male Student: Infinity?
Professor: Well, that's not a number. So that's the singularity. If there was some place where the volume was 0, there would be a singularity there.
Infinity isn't a number, so the model doesn't predict what it would be. So that's where the concept comes from in physics. If you have a situation that doesn't fit the mathematical model and something weird happens like you don't get a numerical result- you've heard you're not allowed to divide by 0 because they don't know what the answer is going to be.
Here's... you probably think it is infinity and I'm just kidding around. But supposing you have 10 divided by 1, then we'll make 1 approach 0. It gets closer to 0. But if I have 10 over -1, then have it approach 0, the whole thing gets close to what? 10 over something negative as this gets closer to 0... it gets more negative.
So it ends up being minus infinity. So it depends on whether you think it's a -1 increasing to 0 or a 1 decreasing to 0
So that's why, you know, if we had a point in space with 0 volume, the density would be a singularity.
So it's thought that there might.... okay, so the same thing is thought to occur with increasing technology where if you try to calculate the resulting power of technology you get some non-numerical... you know, maybe you'd think it was infinity or something the math wouldn't give you an answer.
Well, here's another example. This would be like an economic singularity. Let's suppose... okay. Labor productivity is increasing a little bit every year. How many people knew that? Anyone learn that in economics? What does that mean that labor productivity is improving?
Female Student: We're producing more with less labor.
Professor: Right. Less labor produces more. Just like you said. So what happens if that process continues to the point where the required labor to produce something reaches 0?
What would it be like? Let's take bread. Let's suppose the amount of labor to produce a loaf of bread reaches 0. What will happen? What would happen if the labor required to produce a loaf of bread was 0?
Male Student: No more hunger. Free bread.
Professor: Haha, okay.
Male Student: You could make more profit off of it.
Professor: Okay, that's an interesting thought. Supposing you owned a bread factor and you could produce unlimited bread for 0 labor.
Male Student: But if everyone could do that with 0 labor, everyone would do it.
Male Student: No, the companies would get together and set the price.
Professor: Well, you might hope that sooner or later someone would say "I'm going to break the fixed price and sell more Bread." Assuming prices weren't fixed, there would be an unlimited supply of bread. What if the required labor of producing many things reached 0? Then there'd be an unlimited amount of lots of things.
For example, more powerful cell phones are getting cheaper. If the labor required to make a computer was 0, everyone would have one.
Male Student: Would that cause an economic collapse?
Professor: Well, it might, but it would be an economic singularity because we don't know what would happen. Economic theory would fall apart because it requires that valuable things cost something to make, and we're saying everything would be free.
So you know, conventional economic theory would fall apart.
Male Student: The bread would be free, but they'd get you at the shipping.
Professor: Let's suppose the transportation became 0 too.
Male Student: I give up.
Professor: So can anyone think of a way labor might become free?
Male Student: Socialism.
Professor: Well, even in socialist countries you have to pay for stuff.
Male Student: Robots would do everything.
Professor: Okay, let's suppose that. If a robot was as smart as a person, robots could do all the work and labor would be free. Do you think that could really happen?
Male Student: Yes. It's already happening.
Professor: Robots are doing more and more things.
Male Student: That's why companies are moving back to the US- they don't have to use people.
Professor: So we really could have an economic singularity if robots could do the same work people could do. So if you think a robot could do the same work a person could do, then that cost of labor can become nearly or actually free, which means more could be produced.
So we could have an economic singularity on our horizon. I think that's what they mean by a technological singularity- technology becomes so great that things like this happen.
So things could get really weird if robots could be made to replace people in terms of making things.
So I think there's a thought that the technological singularity would produce an economic singularity.
The AI singularity refers to robots building even smarter robots provided we can build one smarter than us.
Here's another kind of singularity- lifespan singularity.
One of the things Kurzweil talked about was the hope for immortaility. What's immortality?
Male Student: Living forever.
Professor: Okay. Yeah. That's exactly what it was. For thousands of years people have been chasing the fountain of youth. There was a myth that it would make you younger if you could find it. People wanted to live forever
And Kurzweil thinks that it's going to happen, and he's not the only one. He's not the leader of the immortality movement, but he's hoping for it.
Do you think that would be a good thing? Immortality if it became available?
Male Student: Imagine social security.
Professor: And that's a problem. You're right. Population would start to increase dramatically, and if people could still retire at age 65, pretty soon everyone would be retired. Or at least in 65 years, and then that would be a problem.
Of course, if there's free labor from robots, than maybe it wouldn't be a problem.
I have a few more words about the lifespan singularity.
Here's a thought. This book- the cover was mentioned in the movie. [On board.]
That's sort of what the movie was inspired by.
I mentioned that he founded a university- a place where people can go to take courses called Singularity University.
Let's talk more about that lifespan singularity.
So I mentioned Kurzewil is not the main guy in this movement. It's associated with Aubrey de Grey- the chief proponent of immortality. This is his idea: his point is that medical science is improving every year- all the time.
It'll help keep us healthy longer. The idea is that every year, your expected life expectancy decreases. He's saying let's suppose that... I'm going to simplify for a moment.
Let's suppose your life expectancy is 60 more years, then in 1 year it's 59, then two years it's 58. He's saying medical technology will improve, so instead of decreasing by 1 year each year, your life expectancy will only decrease by a part of a year.
If you age a year, your life expectancy might eventually stay the same because of medical technology.
They call this escape velocity. The amount of time your life expectancy decreases is improved by medical advances. If you could reverse that, you could live forever. That's the concept.
Male Student: Have you ever seen the move In Time? Basically, all currency is in time, so on your wrist it'd have the time. Once your time ran out, you'd died. So the rich people were basically immortal, but then the lower workers, whenever they punched out from work it'd just add time to their watch. If they bought food or whatever, it would take time. It was a really interesting concept. This guy meets the rich people and takes their time and infiltrates their lives.
Professor: That is interesting. One of the concerns about using medical technology to improve life span is that it could be only accessible to people who have money.
I think a lot of people think that would be a problem. What do you think? You shouldn't be able to buy everything, right? Seems unfair.
Male Student: That's what happened in the movie. People were making less time than the time they were working because the rich people were controlling it to where they could have all the time and the poor people would just work and die
Professor: That's pretty interesting.
Male Student: Yeah, it was cool.
Professor: Okay, I said I'd simplify it and then I would make it more complicated. So let me do that. Let's suppose you're 95 years old, and your life expectancy is 1 year.
Or 2 years. So you're 95, life expectancy is 2 more years. You wait 2 more years. Will all the 95 year olds be dead/ no. Many will, but some won't be. What will the life expectancy of a 97 year old be? Maybe a year and a half.
So in fact, life expectancy.... calculating life expectancy for a given age is less simple than what I mentioned. It doesn't decrease quite so cleanly, but I want to show you a spreadsheet where I've modeled how things work.
Here's some data I downloaded from the web.
This is life expectancy in the US. For a newborn, life expectancy is 77. For a 21 year old, it's 58.8 years. It's probably lower in Arkansas but higher for college graduates.
Okay. I modeled it as 60 years for a 20 year old. I should have said 58.8.
You can see that for a 25 year old it goes down by 4.7 years. If you get to 30, it's down by 4.7 more years. You're not losing 5 years every 5 years.
But you can see that by the time you get to 100... if you're 95, it's 3 years. And it'll keep decreasing. I tried to make a model that would be reasonable to do and so on, and I came up with this. Let's look at this column. You have age, life expectancy which is decreasing not by 1 year per year, but by how much? 0.8.
So that's 0.8 years per year you lose. I realize this is a simplification of reality, but the results are not too out of whack. You can see the percent of people surviving gets lower as you get oldie are I have 3% of people left at 95. So it's a simplified model, but it kind of captures the fact that most people don't make it to 95.
Alright, well what we need to do is do something about that 0.8 years per year.
We need to... to reach this escape velocity, we need to use medical technology to reduce this 0.8 to 0 so that age 21 your expectancy is 60 and it stays at 60. No matter how old you were. If your life expectancy was 60 years, does that mean everyone is immortal? Here, we're getting to where people are mistaken about their ideas.
So if life expectancy was 60 years this year, and then 10 years later it's still 60 years, are you going to live forever?
Male Student: No
Professor: Why not?
Male Student: You could get hit by cars and stuff.
Professor: Yeah. That means some people will make it less than or more than 60 years. Let me make it more intuitive. Let's suppose life expectancy was 2 years no matter how old you were.
If a year later it was still 2 years, some people still wouldn't make it two years, but some would make it more than two years.
So here, this spreadsheet models what will happen if people reach escape velocity and everyones life expectancy was 60 years. Some people still die from year to year, and... okay. We're looking at these two columns.
We're looking at those two and this one with the ages.
What this shows is how many people die each year will be... if I figure half make it less than 60 years, then half will make it more than 60.
So that's 100% of the people minus 1/120 of that 100% who don't make it through that year. So I subtracted 1/120 each time.
Even if we reach escape velocity, your life expectancy is still 60 years, but by the time we consider 95 year olds, 53% of the people are still alive. That's a lot better than 3%, but it's still only about half. No matter how old you are- only about 1/2 of the people will make it to 95.
Well, that kind of tells you something. No matter how good medical care would be, you could always get hit by a bus. Or there's always, you know- it's a matter of averages and some do better and some do worse, so there's no guarantee about immortality. If you keep this model, how many people make it to... I think older person known to exist was about 120.
But if life expectancy is 60 no matter how old you are, 43% will make it. But if you keep going.
How long will it take before practically no one is left alive? 0.5%. I didn't change the spreadsheet.
If it's 60 years no matter what, then after 353 years, 6% of people would be left. So even if we reach escape velocity, no one is going to last forever. 6% would last 353 years, but that's not a lot.
So my point here is that my idea here is over simplified, and even if we reach escape velocity, you won't live forever. If life expectancy never decreased, the average age at death would still be 102-103.
Alright, let's look at a partial... you know, maybe we're not going to reach escape velocity, but some people say that we're gaining about a quarter of a year every year from medical technology. Life expectancy gets a quarter of a year better than it would have been without medical technology. That's what these two columns model.
So here, what we're gaining over the original column- we're gaining a quarter of a year every year. So it decreases a quarter of a year less than it would have normally.
Then let's see what happens. Let's look at the 50% mark. Right about here, which would be 78. The average person would live to be 78.
Which in fact is about right.
Okay, so if you read up on de Grey, you'll hear about this concept of escape velocity, but the important lesson is that this doesn't mean automatic immortality for everybody.
Okay, so the escape velocity concept. Does it work? Not really. Any comments or questions about that? Okay, let me show you a few graphs.
Actually, I just want to... we don't have time for everything, but there's more good stuff.
Let's go back to the economic singularity. Wouldn't it be nice if the amount of labor to produce a car was 0? Cars would be really cheap. And everyone could have a car.
What would happen if a factory was so automated that it didn't have to have any people. And they're trying to do that. They're doing more and more with fewer people. There's a pasta factory near the town I used to live in. I heard that it only employed like 10 people in the whole factory.
If a whole factory could exist without a single person, then the factory could produce output and get money, buy more materials, and it could be self perpetuating. If it was owned by a person who could take the profits, that would make the output cost something. But if it was owned by the government, it would essentially produce stuff for free. That would be pretty cool.
And anything you could make in a factory that didn't require labor would be almost free. You'd still have to pay for raw materials, but the output would be limited by the cost of the raw materials. But if those raw materials were produced by free labor, then they'd be free too.
So if a given factory had no people working in it, then not everything would be free, but if all production was done by robots, everything would be free.
So that's what the singularitarians. Looking forward to- a day when robots would do anything.
Okay, let's look at some graphs. I'm going to show you some kurzweil graphs. He has some good ones
This will show you some of the specific technologies that are increasing exponentially and may help bring about the singularity. Some of these are redundant, but let's look at a few and see what we can come up with.
Countdown to the singularity.
Here, he's showing creation of life 1 billion years ago. These are all sort of major points in the history of the earth.
[Teacher reading: [On board.]
Agriculture- 10,000 years.
Telephone- 100 years ago
I think the idea here is that important things are happening faster and faster.
Before, things took billions of years. But to produce the... so I guess the idea is that these dots are equidistant. Telephone, electricity and radio is the same time span.
But you see, since this is a logarithmic scale, the distances represent different time frame. Things are happening faster. Do you agree? A lot of people don't like this graph.
Male Student: What is homo sapien sapiens?
Professor: Yeah, okay. There used to be different species that were similar to humans but not quite humans. There were dozens that died off. We're the only one left. They say that most modern humans have a couple percent of neanderthal genes. So we're all part neanderthal. Besides from that wrinkle, this is our variety of homo sapiens.
There were other beings that were similar to us, but we look a little different.
From a biological standpoint, this is a variety, and this is a species. Modern race categories are not covered here at all. We're all the same variety from a genetic standpoint.
Any other questions or disagreements or agreements about this graph?
So do you agree that things are happening faster and faster? This is intended to show how the singularity is happening
Soon, the intervals of time will be really short.
Some people say that well, great. The dots are about equally spaced, and yes this is billions of years, and this is only 100s. But if you look at these dots, what are they showing? Is it really correct to say the development of agriculture and city states is of the same magnitude of going from one celled organisms to reptiles or from apes that walk upright to human beings? Is that transition the same magnitude of going from radios to computers? Many people would say these are much bigger changes than these.
The claim that things are accelerating requires the assumption that all of these represent identically huge changes.
But inventing the computer was a big deal, but not as big as the creation of reptiles.
The inventing of printing- that was a big one. It made books happen. Of course, now we're moving away from books. But it wasn't as big a deal as the evolution of primates
We all have the same body shape to some degree. We don't look like dogs on something. The complaint people have with this graph is that they would claim that yes, time is accelerating on this axis, but the these dots represent... you know, they're more important here than they are here, and that makes this graph meaningless.
Well, what do you think? Are the dots equally important or not?
Anybody think they're not?
Male Student: You can't get to the next one without the one before one
Professor: True, but is the difference between the invention of printing and the industrial revolution the same as the evolution of the first mammals and the existence of the first primates? I have my doubts. I don't think they're the same.
But time will tell, because if these were all equally spaced, it wouldn't take much longer before things started happening once a month, then once a day, then once an hour. I don't think that will happen. This is one of kurzweil's favorite graphs, but I think it has a few problems.
Let's see if we can find another graph.
I can't even tell what the graph is until I click it.
That looked like a good graph. Let's see. Okay this shows an exponential curve.
The date is on the bottom.
You probably can't read it. It says 1920, 1940, 1960. So 20 year increments.
This was made in 2011
So this shows the power of computers- some computers, not all computers.
And what it indicates if we look at it- this shows that the current computer... I don't know what they're talkigna but.
Male Student: That's a graphics card.
Professor: This must be in personal computers or average kind of powered computers. The power of this computer was lease than the power of the brain of a mouse. The mouse is here- 2011 we were only about up to here.
It's the power of computers- probably millions of insturctions per second. This suggests that we'll have computers that have the processing power of a mouse's brain in that much.... let's see what the date is.
Male Student: It's 2015.
Professor: Oh, right. That's not too far off.
Alright, well, why stop at a mouse? They're pretty dumb anyway. Let's talk about a person- a human brain. We'll have a computer that may not be as intelligent, but will have the processing power of a human brain around 2023, according to this graph. Things get weirder after this.
Let's say there are 9 billion people in the world. So that's the processing power of this times 9 billion. They think that will happen in 2045.
So haven't a computer with processing power of one brain is one thing, but if we do one brain, we can do 10 billion brains in 22 more years
Down here, in 2000 we had computers way less smarter than mice. That's pretty weird to even imagine that in 2045 you'll be around, and according to this graph you'll be able to go buy a computer with the processing power of all brains in the human race. That's weird, right? I don't know what we'll have those computers doing. My personal feeling is that we'll have trouble programming them. Kurzweil isn't so pessimistic. They think life will be way different. Hope I make it that long. You guys probably will.
You'll probably be about my age now. You'll be able to look at this graph and say "well, it did/ or didn't happen.
Are you convinced? Do you think it'll really be like this?
How many people think it might well be like this? How many people are pessimistic and think it's total BS?
Male Student: I think it'll start sloping off, you know what I mean?
Professor: You hit on an interesting thing. These people who believe in the singularity think the curve has been exponential, so it'll stay exponential. But remember in class, we talked about how long term these graphs level off. We don't know where that point is, but to say it'll go off exponentially forever neglects this fact.
My opinion is like yours. I don't see a necessity to limit the power of computers. I will say that most exponentials don't go up forever. They do tend to level off. If you remember back to the beginning of the semester, what did we say if we happened to this longer term. What does it do?
Male Student: Doesn't it go back down?
Professor: Yeah, we talked about plateau curves. The number of people with land line telephones went up exponentially, but now we're probably around there. No more land lines. People in underdeveloped countries are now installing cell phone towers. Does anyone here not own a land line? Some people?
Male Student: My parents still do.
Professor: So you're basically living with a cell phone. When I brought my daughter to college, there was a land line outside of her room, and I told her we should get her a phone. Then one of her roommates installed a phone but then they got all of these collection calls for the person who used to live there.
So next time we'll continue and talk about something else.