#
I had a strange dream the other night. I saw my wife in a dream. Now, this is unusual
#
by itself. Who dreams about their spouse? Anyway, in my dream, I was in an amorous mood
#
and I went into my bedroom hoping to interest my lovely wife in some playful interaction.
#
She was curled up on the bed, laughing to herself in a way I only see when she is laughing
#
at my jokes. She had an electrode plugged into her brain. I opened my mouth to say something
#
and she held up her hand and shushed me.
#
I don't need to talk to you anymore, she said. There is no longer any need for any actual
#
physical interaction between us.
#
What are you talking about? I said. How can you say such a thing? Don't you love me anymore?
#
Oh, I love you very much, she said. But now, I have an enhanced version of you. I uploaded
#
your brain on this device a week ago when you were sleeping and now there is an AI version
#
of you here which makes even better conversation than you do. It has all the good things about
#
you, your wit, your knowledge. But none of the bad stuff like your inflated sense of
#
self-importance, your lack of empathy, your short attention span. You are Amit 1.0. This
#
Oh damn, I had always feared this day would come. But that's just conversation, I said.
#
What about the physical stuff? What about cuddling? Ha! She laughed. That's also sorted.
#
See Amit, people are a technology to make you feel a certain way. When you hug me, oxytocin
#
floods through me. Well, now I can replicate the firing of neurons that leads to that without
#
actually having to hug an actual human. In fact, over the last few weeks, I have saved
#
all my experiences with you into the hard drive right here. Now I can replay them any
#
time I want and even tweak some details. I don't need you at all.
#
I was very sad when I heard this. I started pouting. But as my natural resting face is
#
a pout, I must have looked the same.
#
Okay fine, I said. I understand. I apologize for intruding. I'll just go and record next
#
week's podcast now. You don't need to, she said. It's done. I'm listening to it now.
#
You talk about me in your intro. How sweet.
#
Welcome to The Scene and the Unseen, our weekly podcast on economics, politics and behavioral
#
science. Please welcome your host, Amit Barma.
#
Welcome to The Scene and the Unseen. In today's episode, I'm going to be talking about artificial
#
intelligence which is changing the world around us mostly in good ways. Unlike some alarmists,
#
I have nothing but positive feelings about AI. I have two guests on the show today who
#
share my optimism but are also aware of both positive and negative unseen effects of artificial
#
Ramis Naam is a renowned futurist and award-winning science fiction novelist. He works at the
#
Singularity Institute and I urge you to Google him and read some of his essays on AI. Also
#
on the show today is Pawan Srinath, my colleague at the magazine, Pragati, who has studied
#
some of the public policy implications of AI, such as universal basic income. Ramis
#
and Pawan, welcome to The Scene and the Unseen.
#
Ramis, you've written a lot about artificial intelligence and I've been reading a lot of
#
your stuff and it's amazing. So tell me something, especially in India, there's a fear that people
#
have about artificial intelligence that it's going to decimate jobs, especially in the
#
service industry, for example. And like at one level, the scene effect of artificial
#
intelligence creates enormous value and it might seem the unseen effect is a job loss.
#
But to a lot of us here in India, the job loss is also sort of the scene effect. What
#
are we missing? What's the bigger picture here?
#
Well, I think the dialogue is pretty far ahead in India if you're already talking about job
#
loss from automation. I do think we should just acknowledge for a moment that the big
#
obvious effect of AI and digital technology in general is lots of wealth creation and
#
lots of deflation of the cost of services that were once impossible. We all have more
#
access to information than any US president before Bill Clinton, let's say, basically
#
for free. So that's amazing. Amazing, right? Now, will AI destroy jobs? Who knows? I would
#
say it will destroy some jobs for sure. Will it overall have a macro effect of destroying
#
jobs? I am cautious about predicting that because it's been predicted so many times
#
that AI would destroy jobs or that automation would destroy jobs in a macro sense and it
#
hasn't ever been right, at least not for the long term. I'll give you an example of that.
#
I will answer your question eventually. I'll give you an example of that is Ned Ludd. The
#
Luddites are named after him. Ned Ludd may or may not have actually existed. We don't
#
really know if he was a real person or not. Fake news.
#
Yeah, Ned Ludd might be fake news. But he, apocryphally, destroyed artificial weaving
#
machines because he thought that they would take jobs from weavers and textile workers.
#
But the invention of those machines actually increased demand for textiles so much that
#
more people were employed in the industry ten years later than had been before they
#
were around because they brought down the price of textiles that made clothing much
#
more affordable and that spurred demand. It was a positive some sort of thing.
#
Okay, so we have some humility about predicting the future. But let's imagine that it does
#
destroy some jobs. The example you and I were talking about last night is Norman Borlaug.
#
Norman Borlaug started the Green Revolution in the 1940s. He bred better wheat and then
#
better rice that doubled crop yields around the world, helped save India from massive
#
famine, saved Mexico from massive famine. They say he fed a billion people, saved a
#
billion lives, maybe more than anyone in history. Borlaug grew up in sort of a poor town in
#
Iowa and the US. He had no electricity, no running water. And the reason he is who he
#
is today is that he would have grown up to just be a farmer following in his parents'
#
footsteps, but his family was able to get a Ford mechanical tractor. And that Ford tractor
#
destroyed Borlaug's job of being a menial laborer on the farm, but that let him go to
#
high school, not even university, high school. And that led to him having the skills to combine
#
new ideas to produce these better crops that feed us all.
#
That changed the world.
#
That changed the world.
#
And made him a hero in India. So, you know, I'm always an evangelist for technology. To
#
me, the unseen effect of more technology is always positive. You know, you might see some
#
immediate job loss in the short term, if at all, but so much extra value is created that
#
it goes back into the economy and the world is better off than it was before. It's a positive
#
sum game. What I don't have answers to and what people often ask me is in the specific
#
Indian context where we are, where as a nation we're growing younger and younger, we have
#
a million people coming into the workforce every month and there simply aren't enough
#
jobs getting generated for them. And what I see happening is that a lot of our service
#
industry is low end kind of services, which can easily be replaced by AI and has already
#
started happening. And plus, while we never had a manufacturing industry because of our
#
labor laws and whatever, that is a non-starter now simply because automation won't allow
#
it to happen. So now I understand that in the long term, great things will happen. And
#
I also, by your point about having humility about predicting the future, because the future
#
is of course full of unknown unknowns, as it were in this case. So no one can possibly,
#
no one could have predicted at the time Ballock's parents bought a tractor that it would lead
#
to the green revolution. That was an unknown unknown. But before the long term comes, we
#
got to deal with the short term sort of social unrest that is likely the demand on policy
#
makers to somehow ameliorate this is likely. And when people ask me all this, I really
#
don't have any answers anymore.
#
Yeah. It is complex and there are parts of it that are really scary. We know whether
#
or not the macro job destruction happens. We know that micro job destruction will happen.
#
In the US, we talk a lot about truck drivers. In most US states, truck driver is the most
#
common profession. But those might be automated away in the next 10 years, even if it goes
#
very slowly, it might be 20 years. So what do they go do? It's about 3 million people
#
in the US. About 1% of the US population drives a vehicle for a living. And I see a lot of
#
people in India that do that as well. So maybe, and I'm optimistic about this, the total number
#
of new jobs created by technology will be much larger than the number that are destroyed.
#
But still, what do you do with those people? I hope for a couple of things. I hope one
#
that technology that's disrupting driving or the disrupted music or newspapers might
#
also disrupt education. So MIT, one of the world's top universities has said they will
#
make all of their curriculum available for free. We have AI in video games that knows
#
how skilled you are and plays just hard enough for it to be an engaging experience for you
#
and in fact makes it addictive. Could we have AI tutors in your phone, in your tablet that
#
have access to all of this curriculum that know what questions you're getting right and
#
wrong and tailor the education for you in a way that no human teacher with 30 or 40
#
kids could do? That sounds like science fiction, right? But it was science fiction for a robot
#
to be driving a car just 10 years ago. So if we do that, is there a chance we can take
#
children throughout the world that maybe don't have good schooling and give them amazing
#
schooling, far better than now, for pennies, billions of them, and even take adults who
#
are taxi drivers? They might not have the same ability to learn new skills, but I'll
#
bet we can retrain them into some new jobs if we deploy this sort of technology for them.
#
So let me play devil's advocate for a moment. A government policymaker, say 10 years down
#
the line in India, might well say that, look, I buy that technology as a whole is a great
#
thing, but I can approach different aspects of it differently. For example, if truckers
#
across the country are losing jobs, I can ban self-driving cars or tax them or whatever,
#
and I can let the education happen. It's not as if I have to have the same approach
#
to a disruptive technology and to technology which clearly does the kind, and I absolutely
#
share your optimism on the online education bit. So at a policy level, if it's approached
#
like that, and even in the popular imagination, I mean, populism dominates the world today.
#
I don't see it for the same reason. For example, some of Trump's victory is definitely due
#
to job loss, which he attributes to, A, jobs being shipped overseas and B, immigrants coming
#
and taking the jobs, and a significant chunk of them are because of automation. How long
#
before the angle goes in that direction? I do think unrest is a real issue. I do think
#
people see their way of life changing and they have economic anxiety. There's sort of
#
a hollowing out of the blue collar workforce in the US, the manual laborers that used to
#
work in manufacturing and so on. Service jobs are vulnerable, but they have not been hurt
#
as much so far. I think India does have some vulnerability there. So you have to figure
#
out how to deal with that. And I think part two is you have to have a social safety net.
#
You have to have some way to say, okay, if your job has been destroyed, we are going
#
to take care of you for a while with incentives for you to learn something new that is valuable
#
to society. That that should be the number one thing that we ask someone to do if we
#
are taking care of them when they're unemployed, is we should ask them to learn fundamentally.
#
Parvind, Ram has mentioned the social safety net and you've been reading and thinking and
#
writing about universal basic income, for example, among other things. What are your
#
thoughts? I think universal basic income seems to be the latest idea in welfare that has
#
picked up speed. The idea has been around for a while. Various flavors of it have existed.
#
There's the idea of a negative income tax, for example. And I think it's good that this
#
discussion is happening in the United States because I think they are prosperous enough
#
to have that conversation. In India, even if you take the pie, you redistribute it,
#
you get rid of administrative costs. You get nothing. You get 3000, 4000 rupees per person
#
per year. If pushed, maybe 20,000 rupees per person per year. And when you're providing
#
that as cash and you don't have public goods that are available, then your efficiency in
#
using that cash is very limited. You don't have a road to travel on. You have the money
#
to buy a scooter now, but that's not going to help. So I think our conversations have
#
to be a little different. And I want to ask a question on this. I think one of the ways
#
to ameliorate economic anxiety on this is when people can think of easy first entry
#
jobs that can be created in this new space. If I'm looking at what's happening in India
#
over the last five, 10 years with e-commerce becoming a big thing, with new startups sort
#
of achieving scale, is that you have a lot of these jobs that people can quickly get
#
into. You're a driver. You're a delivery guy. You are in many other places and these are
#
being created rapidly. So one of the reasons why startups have worked in India so far is
#
while they have hit at disrupted old businesses, they've disrupted the black and yellow taxi
#
cabs in Mumbai, but they have also created these jobs. So there's a new constituency
#
of people who are batting for them. So in AI, do you see any opportunities for such
#
jobs? I think we do. And I would say that here's a slightly more sophisticated way to
#
think about job destruction. Most of the time automation does not destroy a whole job. What
#
it does is that a job is a basket of tasks. Let's say your job has 10 tasks. Let's say
#
you are a delivery man. Some of your task is you walk in to the restaurant and you get
#
the food. You walk back out. You know the address. A bunch of your job is you drive
#
there. And then another part of your job is you get out. You walk up to the house. You
#
knock. You give the person their food. Collect the money. Well, technology might automate
#
away the driving part, but that last, not even the last mile, the last hundred meters,
#
it's actually quite difficult to automate away. And drones, we can have fantasies about
#
the taco drone all we want, but it just doesn't make a lot of economic sense for some time.
#
And we see that in a variety of other things. In white collar work and service work, we
#
see the same thing. So I'll give you an example of something not in the tech industry. Being
#
a lawyer in the US, one of the large things that you do in a big complicated case is discovery.
#
After this discovery, it's reading through thousands or hundreds of thousands or millions
#
of documents to find something interesting. So now we have AI that can automate, can semi-automate,
#
can make that part of the job the most boring, least fun, least intellectually demanding
#
part of the job, much more streamlined, and it can free up the attorney to do the most
#
important parts, the negotiation, the arguing in court and so on. So I actually, I'm not
#
sure that in many of these cases that we'll see people just go away. If the truckers that
#
I talked about, if that trucking job is automated, you still have someone that needs to refuel
#
that vehicle. You still have the people that need to load and unload it. Some of that will
#
be automated, not all of it. You are reducing the cost of shipping goods. That will probably
#
in turn increase the demand for shipping goods. And the way the textile demand went up when
#
the price went down, that in turn will create ancillary jobs. So that's what I see. And
#
I can't tell you exactly what that looks like, but I think if, for instance, if India had
#
fully autonomous vehicles, every single one, A, the traffic would move five times faster,
#
and what economic surplus does that have for everyone in India? And B, will that actually
#
create more demand for delivery services that employ a human in some way? Maybe the human
#
has to do four other tasks during the drive, but he's still going to do something.
#
Right. Absolutely. I mean, I completely get this exact analogy just with Uber and Ola
#
in India. If you had to be a taxi driver in an Indian city, say 15 years ago, you needed
#
to know the city first, right? You needed a geographic map of the city. You needed a
#
mental map of this. And now, because Google Maps has automated that, any first time driver,
#
so long as they know driving and they know the rules of the road reasonably well, they
#
can just get on the road.
#
This actually made it easier for people to become drivers and therefore increase jobs
#
So, like you said, if certain tasks that require extensive experience or training become obvious
#
with the use of technology, maybe some jobs might increase in certain sectors.
#
Yeah. Maybe the future of meal delivery is going to be that it's a mobile kitchen and
#
the person is, the work they're doing is they're cooking inside while it's driving automatically
#
from each place to each place. I don't know. I'd like to come back to the universal basic
#
income though, because when I do the math, I just do the math in the back of the envelope,
#
what I come to the conclusion of is we should talk about basic income and not universal
#
basic income, because if you target it at the bottom 20% of society, let's say, you
#
get five times as many rupees that you can spend on those people. So, I think we should
#
think of it as a safety net, something that phases out gradually and slowly with income
#
rather than being universal. There's various arguments for universal that gets more politically
#
popular if everyone gets some, but it's just not as efficient. I don't need that income
#
myself. I would rather that a poor person gets more than I do.
#
Right. I think the argument in favor of a universal anything is that the cost of targeting
#
might and the challenges of mis-targeting might outweigh the cost of universalization
#
and I think with technology that might be proving wrong. We'll need to see how Aadhaar
#
and other things in India can be deployed meaningfully to make targeting successful.
#
What Aadhaar will do is it will provide universal surveillance.
#
So while we're looking at that angle, I think the original purpose of the Aadhaar was how
#
do we manage horribly bloated subsidies in India?
#
Fair enough. How do we get the targeting right? How do we make sure the right guy is...
#
I fully agree with you.
#
So let me end by asking a question of what could either be a utopian or a dystopian future.
#
Now typically we imagine that, you know, whenever new technology comes, it might cause short-term
#
micro job losses, but it creates value and that value goes back into the economy and
#
that creates more jobs. Now what happens in a future scenario where artificial intelligence
#
and automation together can essentially satisfy every human need, right? And in which case
#
whatever value is created goes back into AI and automation because any need that any human
#
can have is possibly satisfied. Now the dystopian vision is that, my God, there are no jobs
#
and what are people going to do? And we need to protect the people. And the utopian vision
#
is that everything will be so cheap because productivity is so high that people don't
#
need jobs. Where's the balance?
#
So I think people do like having a purpose in life. So we have to worry about that and
#
also expect that people will look for purpose in life. But I also become very, very cautious
#
about predictions like this. John Maynard Keynes, arguably the greatest economist of
#
the 20th century, predicted that in the USA by today, by actually like 2000 I think, people
#
would work an eight-hour work week because he said per capita income grows at 2% per
#
annum and so by 2000 everyone should be able to work eight hours and have a good quality
#
of life. I wish he meant a 1920s level quality of life. And in the US, maybe you could work
#
eight hours and have a 1920s quality of life, but no one wants to. Almost. A few people
#
do. Human needs expand. I'll tell you just from my country, because I know the stats,
#
the living space per capita in the US, number of square meters per capita has gone up by
#
a factor of three since 1970. Why? Because people, given the option like space, the number
#
of miles flown per capita has gone up by a factor of like 20 since 1970. People like
#
to travel. So human needs are not so easily satisfied. Some people would view that as
#
a horrible thing. I view it as a sort of a positive. And maybe we all need to go to Mars.
#
Maybe Elon Musk might think so. Yeah. For now guys, thank you for coming on the scene
#
and the unseen. It was a pleasure talking to you. Thank you. Thank you. Thank you for
#
listening so far into the show. In a future age, this podcast will be uploaded directly
#
to your brain. For now though, we live in the cumbersome physical world where I urge
#
you to head over to your nearby online or offline bookstore and pick up any of the books
#
written by Ramis. His award winning nexus trilogy in particular is a blast to read and
#
very, very thought provoking. Pawan and I both write for Pragati at thinkpragati.com
#
and you can also check out my blog, India Uncut at indiancut.com. Do come back for more
#
next week. Long after humans have become post-human, the scene and the unseen will keep coming
#
at you week after week after week.
#
Next week on the scene and the unseen, Amit Varma will be talking to Pranay Kottasthane
#
about centrally sponsored schemes. For more go to scene unseen dot n. If you enjoyed listening
#
to the scene and the unseen, check out this exciting new podcast from Indus Vox Media
#
called Keeping It Queer. Keeping It Queer is hosted by my friend Navin Narona and he
#
profiles LGBT people from all across the country and some of the stories are really poignant.
#
Check it out on Audioboom or iTunes.
#
Excuse me bhaiya, excuse me. Boli madame. Menu mein kya hai? Menu mein scene unseen
#
hai, podcast hai, on course hai, Cyrus says hai, Marry in India, rediscovery project, empowering
#
series, sex vex hai, IVM likes hai, Simplified hai, Keeping It Queer hai, Tings and Destinations
#
hai, My Neighbor Zuckerberg hai, and The Fine Garage hai. Aapko kya chahiye hai? Ek baar
