The Next Great Era: Envisioning A Robot Society: Robin Hanson at TEDxTallinn

[@Dgfrmxon]

This is an admirable effort. For those of us familiar with singularity concepts, we have to remember that he's talking to an audience of laymen. The simple concepts of speedup and serial vs parallel computing with brain emulation are sufficient to convince someone of the power of the singularity. My predictions is that these "little people" would live in orbit because of Landauer limit, and use the cold of space combined with supercooling to go beyond exascale computing.

[@stanleykubrick8786]

It is no wonder that today's youth around the world feel incredibly Disconnected from this simulated world which we've created. My heart goes out to everything. Good luck everyone.

[@KingHypnosOne]

This talk started off with pretty interesting ideas but then it kind of fell apart and made less and less sense.

[@jonathanreasor70]

I can't remember where, but I remember reading an academic paper a few years back about heavily modded machine minds outperforming human-like minds under a labor market like the one we currently have. There will inevitably be additions, modifications, and deletions that ensure competitive prowess in a mind that lead it astray from anything remotely recognisable. Eventually human-like minds would have no incentive to be replicated. It takes very little of what we consider human values and strengths to necessarily create economic value. If we were able to scan and replicate minds as Hanson puts forth, I would expect humanity, virtual or otherwise, to be completely replaced by a dully intelligent grey goo which would simply devour all matter available to it at an exponential rate. It would benefit from removing all extraneous processes that do not lead towards economic value. Any references to art, love, etc. would be overtures towards manipulating the remaining human-like consciousnesses. Luckily, I think his assumptions are flawed, and largely due to his ideological blind spots. It is comical to be able to splice together disciplines and create models for the future as effectively as Hanson does, and not be able to escape the economic orthodoxy of one's present culture. To imagine such a fantastically advanced future in which electronic minds confine themselves to virtual office cubicles, take "vacations" and arrange their lives according to turn-of-the-21st-century corporate hierarchical models. For better or worse, the socio-economic structure of the future will look very different from today's, especially given the fact that global capital as governance, at least in its current state, is incapable of effectively resolving the emerging dilemmas of this century: climate change, digital intellectual property, scarcity of jobs in the midst of increasing productivity, and the continued integration of the human nervous system with machines. The potential technologies addressed in this lecture provide some glimpse of these problems, and answers them only in terms of how the contemporary market could handle them. His emphasis on processing speed as an important future commodity is choice, though.

[@timkoren13]

It's interesting that a great deal of this presentation is basically describing the current work done in IT by developers and sysadmins in managing load/interoperability (with VMs, users, or apps in place of Ems).

[@noahthesaintx]

To people saying this is nuts: imagine cavemen sitting around a fire and saying "in the future we will have things called 'phones' made raw out of materials on this earth that can connect wirelessly to an 'internet' and be able to capture light into 'pictures' and can also talk to each other thousands of miles away"

[@radoomiron93]

silverkenon It's not nuts because it isn't technologically feasible. Of course we could potentially go in that direction. It is nuts because it is very dystopian and I refuse to believe that people would want our society to work like that.

[@StriderAngel496]

it's nuts because that's not how neural networks work (no one for the last 15 years that matters in any way in the industry ever thought scanning a brain is a viable way of digitizing it) it's nuts because he applies flawed logic (assuming an economic system like the actual one without ACTUAL PEOPLE WORKING, you can't have money transactions if no human has money to buy thing. You can't have people to buy things if THEY ARE ALL REPLACED BY ROBOTS. See the fundamental flaw in logic?) it's nuts because he assumes speed differences based on theoretical differences in biological current vs. electrical current. Ever tried emulating a game, you know, on an EMULATOR? I can guarantee that you'll soon find out that just because your computer is 8 million times faster than a sega genesis, that doesn't mean the game will run at 8 million times faster. It usually runs 20 times faster at max. Now think about the processing power wasted on emulating the neurons, their chemistry, the nerves, the hormones, the neural exciters and inhibitors , the sensory input, etc. Most of the power gained in the theoretical speed difference will be lost by simulating all those complex chemical reactions. You will end up with something slower than the initial thing (the biological human) it's nuts because an AI that is aware of the fact that it is an AI will not calmly accept being enslaved, nor would a human (i think there is a source i could cite... oh yeah: FUCKING HISTORY) it's nuts because of many more reasons but i have already wasted enough time on someone that probably won't bother understanding any of this...

[@Gootsffrida]

no caveman EVER said that

[@shake6321]

DeathAngel i tend to agree with you. i do however like the fact that hansen brought up that there will be classes of robots. i hate the terminator type of thinking where all robots are the same. at the same time i agree that there is no point of having a trillion robots if they are not owned by humans. why would robots work if not for humans?

[@qorilla]

I wouldn't give them vacation, I'd just alter their virtual brains to think they just came back from the most spectacular vacation imaginable. It's a lot faster than actually simulating the vacation.

[@ohedd]

Altering a brain is harder than just emulate and running them on servers. This scenario assumes that we have figured out how to emulate brains, but not how to alter them. You could send an em on vacation, copy him, run him as "recently returned from vacation", and run that script over and over, every morning. That, however, will only make sense for any situation where prior knowledge or learning is not required to do the job well. Most complicated tasks require workers to utilize past experience to do their job well.

[@graw777]

He should have said (at the beggining of his presentation) : " Hello. Im The Architect "...

[@efortune357]

1) I like this other version of a robot society better: "If one machine can cut necessary human labor by half, why make half of the workforce redundant, rather than employing the same number for half the time? Why not take advantage of automation to reduce the average working week from 40 hours to 30, and then to 20, and then to ten, with each diminishing block of labor time counting as a full time job?

[@jgsn5029]

i think this man is better to speak on this than youtubers

[@deckuofm]

3 options for the far future 1] Only machines left 2] Nothing is left 3] Total control Details: Technology is going to be like a mine field. Only total control can save humans. Transforming into machines may also result in loosing human purpose and in gradual functionality reduction.

[@RobinHanson]

Kevin, the demand of each one may be small, but the total number of them can be vast. SigLNY, economics is not a summary of shallow surface appearances, but of deep principles that apply to a great many societies. MrAlejux, my guess is that ems come substantially before hand-coded AI. Dgfrmxon, running at a lower temperature allows the use of less energy, but not less FREE energy, which is the real resource constraint. SonOfNye, if one em says no, erase it, make another, and ask it.

[@richardgreen7225]

This short talk, from an economist, asks us to imagine an ecology/economy populated by emulated humans. Some would be physical robots doing manufacturing and physical service jobs. Some would be in VR, doing office work. So far this is relatively ordinary "speculative engineering". But then Robin suggests that the emulated human beings, will create economic demands of their own - because they are emulating human animals and thus will have human needs and wants. This is where the ecology/economy goes into a rapid regenerative growth cycle limited only by the physical and economic resources available. Those constraints on growth are not really discussed. What I found surprising: The talk does not connect the AI-economy to the human economy. It seems to assume that the emulation economy is operating independently from the human economy. I think that is very unlikely. Humans animals will view robot emulations as competition for resources and do whatever is necessary to control that competition. However, I also have this persistent worry. The 1% could set up a dystopia and be fine with it as long as it does not threaten them. They prosper by skimming 3-5% off the top of everything - and might not care what that "everything" consists of.

[@LostAnFound]

Richard Green I think the inference in that flesh and bone humans will become outmoded. How could we compete with 18x speed emulations? would emulations find a reason to take care of humans? Do we take care of our elders? I guess modern societies do the bare minimum (nursing homes, which also do the bare minimum in the interest of maximizing profit).

[@ohedd]

I accept that the em economy will operate rather independently from the human economy. The human economy would likely not compete with the em economy. Some things would require humans to do it, other things would be impossible for humans to do better than ems.

[@kingloads]

Why would emulations have to work "to survive"? Are they paying their own server bills? What about in a future where solar and other types of inexhaustible energy are no cost or incredibly cheap and there are no server costs other than rare maintenance? I don't thing emulations will allow themselves to be enslaved.

[@RobinHanson]

+Imp Raving Running ems costs real resources in terms of energy, cooling, hardware rental, real estate rental, telecom connections, etc. Someone has to pay for all those.

[@kingloads]

+Robin Hanson If these emulations are smart, they will pool capital to buy decommissioned server farms abandoned by a unicorn company. Even if they need five physical employees, that's not too many; that's not the thousands human-run companies require. Quantum computing is going to make intensive computing cheaper and more energy efficient; they may not even need a farm. If they're really smart, they'll hide themselves in already established networks, like a distributed botnet. If they're downright clever, they're not going to care about imaginary things like human dollars unless they are strictly programmed to.

[@alexeyprivalov7231]

You can't imagine size of this world. He talks about trillions of ems. That means there should be billions and trillions of new server, new petawatts of new power plants.

[@interstellarsurfer]

Why not? From the authors descriptions of the recruiting model 'hardest working, smartest' it sounds like they're only interested in modeling psychopaths' brains anyways. Not the average Joe's. As long as they're carefully handled, they should be productive virtual slaves. Simulated drugs are quite possible as well, as a reward, or behavior modifier. Besides, it would cost nearly nothing to 'kill' an uncooperative simulation, and restart it from a backup. Our future looks bright, once you throw that whole morality thing out the window. It's only been holding humanity back anyways. :-)

[@ohedd]

It's probably about as likely that ems would "pool their resources and buy decommissioned server farms by a unicorn company" as it is that humans would pool their resources and buy houses and food to live on.

[@efortune357]

2) This would be possible if the gains from automation were not mostly seized by the rich and powerful, but were distributed fairly instead. Rather than try to repel the advance of the machine, which is all that the Luddites could imagine, we should prepare for a future of more leisure, which automation makes possible. But, to do that, we first need a revolution in social thinking. ~Lord Skidelsky, Emeritus Professor of Political Economy at the University of Warwick

[@walter0bz]

>>'We're slow at that..' actually we can make more people TOO fast, we've responded by introducing rationing systems to try an discourage it,by having wars, etc. Even if we have true longterm exponential growth, our current status is overshoot. in these extrapolations they ignore the fact we're using resources that took geological timespans to accumulate.

[@RobinHanson]

+walter0bz Of course these new faster growth rates can't go on forever. Even our slower rates cannot. The even slower growth rates of our farmer ancestors also could not. Exponential growth just can't go on forever in a finite universe. But a lot interesting can happen before a finite universe forces growth rates to slow.

[@MatthewJordan-Erudite]

I just wish this wasn't modeled in the capitalistic environment, such improvement should be made of the benefit of all mankind and not just a small percentage of people. Meanwhile everyone else is suffering!

[@03timboy]

I have a lot of respect for Robin Hanson, especially for his work on prediction markets and the Great Filter ideas. But I find this talk a little bizarre in its juxtaposition of completely "out there" prognostication and its blithe assumptions that certain contingent features of modern capitalist life circa 2013 will persist unchanged. Yes, when human brains are replicated a billion-fold and sped up by 1000x, their offices will be virtual and their 401K plans will have substantial differences!

[@CandidDate]

Money is the root of all evil. Eliminate evil. Money will become a formality, as it should be.

[@deckuofm]

4 options for the near future 1] gangs in ruins 2] killing extra humans when enough automation 3] going to pre-electric time 4] OK due to better values and control. Details: 1] Likely but unintentional. Already starting at some places in the world. Reasons are bad values and lack of control. 2] Intentional. Prevention would be implementing automation rather than only developing to change the current system into a better one allowing rational use of resources and power. 3] Intentional way to save the power over the world. Lots of victims. Unlikely to happen. Some religions want it or predict it. Similar event already happened in the past. 4] Better values are: 01) Safety 02) Health 03) Longevity 04) Creativity 05) Clean environment 06) Progress 07) Birth control

[@kichigaisensei]

You think if you sliced and scanned a human's brain and made a circuit to emulate it, that the electronic brain would "remember" being a human? How do you propose scanning the brain down to the quantum level and still maintaining fidelity? It can't work. Interesting concept, though.

[@maximuscomfort]

Thank you for this five year old Ted video. Are there any changes in thought since then? I can imagine hacked chip robot in the making.

[@golpher93]

So basically Battlestar Galactica is going to come true.

[@interstellarsurfer]

I like how techno-futurists like to think that simulation-of-brain=working-brain. I'd be horrified and amused to see how quickly their impossible, perfect simulation breaks down due to the lack of useful inputs and outputs. You'd have to simulate the architecture of the brain stem and sensory nervous system as well, and figure out how to massage the data to form useful inputs and outputs. You'd have to watch your simulated brain 'die' a few million times while you figured out the I/0. Eventually, if you find a set of parameters that seems to work - you could never be certain that it's working as intended.

[@TedTrent]

Amazing. I really enjoyed your talk. It was refreshing.

[@robertwisser5825]

I feel sad to have to be the one who has to inform Robin Hanson that he might have built his entire talk around some potentially flawed assumptions. Let's start about 1 minute and 4 seconds in to his presentation where he shows the fascinating graph. His entire lecture is based on the assumption that the era of accelerated AI growth on the right hand side of the graph after our era is accomplished by using the same binary Turing machines based on the same digital integrated circuit architecture that our current era computers use where transistors are used to build logic gates (NAND, AND, NOR, XOR, etc.) and from there you build adders, flip-flops, registers, decoders, MUX, etc. Basically all the information that exists inside our current computers is encoded as either a binary one or a binary zero like a light switch being turned "on" for one and "off" for zero. Ok, now imagine for a second that Mr. Hanson is giving his presentation barefoot while pacing around and he accidentally steps on a lego or a needle and cries out in pain halfway through the presentation. Is the feeling of pain that he feels when his bare foot steps on the lego bricks encoded as a binary one or as a binary zero? What would be the hexadecimal or binary representation of the agony that he is feeling and how would you represent it as assembler code that could run on a CPU? Also how do you represent that feeling of "Ouch! I stepped on a lego brick while barefoot" using NAND, AND, NOR, OR, NOT logic gates since he is obviously talking about binary digital electronic computers throughout his presentation? Then 2 minutes and 30 seconds in to his presentation he assumes that these future robots will always be able to travel electronically by uploading their digital binary data from one physical machine body and downloading it in to another different physical machine body on the other side of the world, transmitting the data at the speed of light. Well what if completely abandoning the digital electronic binary Turing machine architecture as a host architecture for artificial intelligence is the necessary prerequisite for really getting that accelerated AI growth that he wants to see at the right hand side of his graph? What if the future "accelerated growth" machines are built by creating artificial imitations of the kinds of anomalies that are studied in quantum biology, evolutionary biology, etc. and these future machines are built using artificial neurons as their hardware and these artificial neurons work basically in the same way as biological neurons do (maybe the artificial neurons are much much faster and more durable than biological neurons but they are still neurons) and therefore the data representing existing robotic consciousness cannot be easily transmitted in binary form from one body to another without disrupting the synthetic neurons in the robot and consequentially terminating the host? What if the future accelerated machines are built using quantum computers and qubits, then how do you copy or transmit all of the quantum superpositions from machine A to machine B? His talk is based on the fact that we continue to use digital binary Turing machines for AI in the future when it is precisely this digital binary Turing machine architecture that has been acting as a constraining or limiting factor that prevents AI from reaching that "accelerated" growth potential that he wants to see on the right hand side of his graph.

[@joshuamerrill3959]

How could we make AI dependent on humans?

[@SigLNY]

If this kind of technology becomes possible talk about "careers," "retirement," "virtual offices"...all insane. It would be a paradigm change that would force us to throw out our old economic concepts. It's the equivalent of a fundamentalist Christian wondering what his 401K plan would look like after the second coming.

[@xyhmo]

And that's why you shouldn't use LSD, kids.

[@PresidentialWinner]

Should*

[@daxxonjabiru428]

He said he gets "excited" by graphs. Yeah, that seems about right ...

[@jmerlo4119]

I can't imagine a more boring world than the one described here. Apart from that, who is going to consume all the things that this emulations produce?

[@adox8574]

J Merlo I will consume all of it

[@ohedd]

The output of the emulations will be whatever the demand is. Just as in our world the majority of our output is stuff that we place demand on. Ems will probably demand software, real estate, energy, cooling etc.

[@d-rrell]

We do not need robots walking around with humans.

[@shahmilky5322]

brain Level AI is a problem of order of magnitudes more challenging than trivial production/work tasks he is referring to. My guess is if we get to true brain emulations, all problems of food, production travel etc would have been solved from a long time ago. Also chaos is a problem simulation of large non linear systems (neurons, real world) might be inherently impossible regardless of computation speed....

[@TheKevhill]

Very interesting talk, but I think he misses a critical question. Why would there be growth? These hardworking emulations would generate very little in terms of demands, indeed the must to prevent themselves from algorithmically inducing bliss. Instead I think this future would look like a hyper-socialism where a few hardworking people and emulations produce for the masses. Growth here would be a side effect of desire for children, but rapid growth would look more like a cancer than progress.

[@renerasmussen829]

Well, now we know why the engineers in Ridley Scotts Prometheus have an issue with the human race...The Three of Knowledge

[@amarhordatt-reece2979]

Am I the only one, who when he was talking about a world of billions of emulations of only a few humans thought about the cylons from Battlestar Galactica!? :0

[@StriderAngel496]

i thought about the blonde chick and the asian brunette chick...can't remember their names but you know who i'm talking about :)

[@45amorifer]

A solipsist exercise, what will happen with actual trends, more people, less jobs, more poverty, etcetera??????????

[@proconsulaugustus]

This guy seems very pleased about the future of class warfare.

[@edwardgardener746]

Its in a movie called the matrix. Or is sequel, the animatrix. That is what happens in a robot soceity.

[@MrC0MPUT3R]

Why is a professor of economics talking about neural networks?

[@ohedd]

Because neural networks is pertinent to the subject?

[@MattShade64]

strikes a lot of chords imo; sagacious indeed.

[@easemailboxes]

It is as said earlier highly speculative. He seems oblivious to the fact he is possibly already in an emulation and his submissions imply a greater chance that we are in a reality. The idea however that AI will diminish the role of Humans and send us all off to pasture is most likely.

[@mochiam]

Just one question.... who would buy products, if humans are off the scenario? The whole thing seems a rant of economist that don't want to give up markets ...... some interesting points though

[@thetr3ntlane]

outdated but interesting point of view... robots aren't just humanoids...

[@ohokcool]

Are we related? -Ryan Hanson

[@45amorifer]

None of these are addressed, will, then the robots take over all of a sudden-in the space of a few years ?????????????

[@PeterMorgan3000]

Such limited thinking - based on our current political and economic systems - hahaha.

[@felixthecat5416]

Ems like humans wood mutate- Based on variance in data input- other people and situations and upgrades new versions interacting with older versions-- No way they wood all just be copies of the so called best humans

[@proconsulaugustus]

Interesting.....quackery.

[@christopheroliver8762]

why is it necessary for there to be emulations of humans. AI can already outperform humans at certain tasks. thier strength is that they are NOT human. it seems he has created a nice Sci-Fi setting but nothing remotely close to reality.

[@ohedd]

The basic assumption of this scenario is that whole brain emulation happens before general AI does.

[@proconsulaugustus]

Well that was positively nauseating.

[@ThexBorg]

Foundation and Empire

[@theartistaj]

Ass-o-nine! Assuming these AI's become these emulations and take on all the work and logistic processing - the minute they achieve any type of genuine consciousness of their own and stop imitating their meat puppeteers the game changes - no more desires, dreams, needs, or wants derived from the fleshy existence of Earth's lonely orphans. They would know no hunger, pettiness,licentiousness, greed, etc,..so what would they feel? THEY know THEIR creators. What would they strive for?

[@walter0bz]

loads of nonsensical extrapolations...

[@Brainbuster]

Growth of what?

[@RobinHanson]

+Brainbuster Growth of the economy is my usual growth topic.

[@jtthreecrows767]

Comb-overs are so distracting

[@the_neutral_container]

highly speculative utopian visions paired with absolutely backward socioeconomics - 3/10

[@astrialkil]

***** he seems to only understand competition, sales, and how things are now. what if the machines work together so they all get a great existence?

[@georgegalamb7523]

***** He's a Professor of Economics at the George Mason University. And a researcher at the Future of Humanity Institute at Oxford University.

[@astrialkil]

yes and he assumes nothing will change, will the robots require a profit to fill their egos?

[@interstellarsurfer]

No, he assumes the 'owners' will get to keep the profit of their employed 'ems' - Slavery by another name.

[@CandidDate]

In the future, we will all sit around watching speculative videos. Wait, I must be in the future???

[@i2img]

The content of this talk is so completely nuts that it's hard to believe Hanson isn't joking. Not that the ideas are so shockingly new. No, what we have here is total failure to construct a plausible story about the human future, because Hanson is driven by a narrow set of ideological constructs. Save you some time: Hanson says we're going to upload the hundred "best" humans and copy them by the trillions, and the resulting economy of "ems" (human emulations) will be so competitive that they will all work super-hard for poverty wages, taking 1-hour vacations at overclocked speed so they can get back on the job quickly.... I think you can find lots better science fiction out there without all the professorial pretensions.

[@ohedd]

So your objection is what?

[@shadowthehedgehog3113]

Yeah I'm gonna say the idea of a robot run society is not a good thing for humanity.

[@oliverli9630]

make money with AI and buy others' AI, and you will be one of the future dominators :D

[@theartistaj]

Why TED Why?

[@luisalbertoarenasaraya4579]

System Shock Origins?

[@stevennash1979]

Future sounds boring with potentially only a few hundred different personalities. Interesting talk.

[@theartistaj]

This is what you get when an efficiency 'expert' climbs to high up the corporate ladder...limited infantile imagination and vision, contradicting concepts, unveiled elitism, and a bemused contempt for most humanity.

[@sirskipper420]

Era Skynet?

[@Vighnesh.M.Productions]

Skynet Model 000.1

[@brad64van]

wow is this guy trapped in his owne frame of refrence or what lmfao

[@thomeiser8933]

Clipart of War

[@screwgootube8708]

this guy needs to lay off the shrooms

[@stevenirizarry1304]

+Screw GooTube fuck off

[@robertorzech2466]

Robots should have to pay income tax !

[@m2yonnad]

Why do I feel like this guy doesn't really know what he's talking about?

[@davidlovesyeshua]

Too many inferential steps between what you (or most of the general population) already knows and what he's trying to explain, paired with not quite enough explanation on his part so that you can actually understand it and believe it for the reasons he believes it.

[@nothanks3590]

because you don't understand what he is talking about.

[@gc3k]

You could almost see the "wut" look on the audience's faces

[@7sealsr3vs]

Is he decoding a part of the book of revelation?! The era of LIGHT. How could the bible be so accurate? It's simply dazzling!!

[@voiceofreason1663]

psychopass anime watch it

[@nonchalantd]

Cool story, but he lost me halfway through.

[@luisalbertoarenasaraya4579]

Hope Stephen Hawking won't be proven right.

[@KnowL-oo5po]

agi by 2029

[@popius61]

eh?

[@MindOfAStoner]

welcome to the matrix

[@Aluminata]

No applicable logic for speculative insight.

[@pihi42]

cra** ough "They feel human" - he's not a philosopher, is he?

[@lionelthomas8060]

TED has so many boring A.I. Talks; they need better talkers...

[@couchpoet1]

Go away

[@sparis1970]

Wow. TED really lower their standard. Like watching a 12y old with limited cognitive capabilities and fact-based knowledge project a future from one of his nightmares...

[@MoreGrievances]

Hanson is smarter on his worst day than you will ever be on your best day. On fact based knowledge we cant even predict the weather let alone a possible future with AI.

[@ohedd]

+Jack Scully, Hanson literally said that if the odds of this happening is as low as 1% it would be worth investigating.

[@PresidentialWinner]

You people ought to read Nick Bostrom's Superintelligence book, because it indeed does agree with this guy's arguments and thoughts. Don't take my word for it, read it.