Tag Archives: Ems

I Still Don’t Get Foom

Back in 2008 my ex-co-blogger Eliezer Yudkowsky and I discussed his “AI foom” concept, a discussion that we recently spun off into a book. I’ve heard for a while that Nick Bostrom was working on a book elaborating related ideas, and this week his Superintelligence was finally available to me to read, via Kindle. I’ve read it now, along with a few dozen reviews I’ve found online. Alas, only the two reviews on GoodReads even mention the big problem I have with one of his main premises, the same problem I’ve had with Yudkowsky’s views. Bostrom hardly mentions the issue in his 300 pages (he’s focused on control issues).

All of which makes it look like I’m the one with the problem; everyone else gets it. Even so, I’m gonna try to explain my problem again, in the hope that someone can explain where I’m going wrong. Here goes.

“Intelligence” just means an ability to do mental/calculation tasks, averaged over many tasks. I’ve always found it plausible that machines will continue to do more kinds of mental tasks better, and eventually be better at pretty much all of them. But what I’ve found it hard to accept is a “local explosion.” This is where a single machine, built by a single project using only a tiny fraction of world resources, goes in a short time (e.g., weeks) from being so weak that it is usually beat by a single human with the usual tools, to so powerful that it easily takes over the entire world. Yes, smarter machines may greatly increase overall economic growth rates, and yes such growth may be uneven. But this degree of unevenness seems implausibly extreme. Let me explain. Continue reading "I Still Don’t Get Foom" »

GD Star Rating
a WordPress rating system
Tagged as: , , ,

Paul Carr Interviews Me

In this episode of the Wow! Signal Podcast. The topic is ems, starting about minute 35, after an interview with Heath Rezabek.

GD Star Rating
a WordPress rating system
Tagged as: ,

First Person Em Shooter

Jesse Galef:

It’s The Matrix meets Braid: a first-person shooter video game “where the time moves only when you move.” You can stare at the bullets streaking toward you as long as you like, but moving to dodge them causes the enemies and bullets to move forward in time as well. The game is called SUPERHOT … it struck me: this might be close to the experience of an emulated brain housed in a regular-sized body.

Jesse asked for my reaction. I said:

Even better would be to let the gamer change the rate at which game-time seems to move, to have a limited gamer-time budget to spend, and to give other non-human game characters a similar ability.

Jesse riffed:

It would be more consistent to add a “mental cycle” budget that ran down at a constant rate from the gamer’s external point of view. I don’t know about you, but I would buy that game! (Even if a multi-player mode would be impossible.)

Let’s consider this in more detail. There’d be two plausible scenarios:

Brain-In-Body Shooter – The em brain stays in a body. Here changing brain speeds would be accomplished by running the same processors faster or slower. In this case, assuming reversible computing hardware, the em brain computing cost for each subjective second would be linear in brain speed; the slower the world around you moved, the more you’d pay per gamer second. This would be an energy cost, to come out of the same energy budget you used to move your body, fire weapons, etc. There would also probably be a heat budget – you’d have some constant rate at which cooling fluids flow to remove heat, and the faster your mind ran the faster heat would accumulate to raise your temperature, and there’d be some limit to the temperature your hardware would tolerate. Being hot might make your body more visible to opponents. It would hard for a video game to model the fact that if your body is destroyed, you don’t remember what happened since your last backup.

Brain-At-Server Shooter – The em brain runs on a server and tele-operates a body. Here switching brain speeds would usually be accomplished by moving the brain to run on more or fewer processors at the server. In this case, em brain computing cost would be directly proportional to subjective seconds, though there may be a switching cost to pay each time you changed mental speeds. This cost would come out of a financial budget of money to pay the server. One might also perhaps allow server processors to temporarily speed up or slow down as with the brain-in-body shooter. There’d be a serious risk of opponents breaking one’s net connection between body and brain, but when your body is destroyed at least you’d remember everything up to that point.

To be able to switch back and forth between these modes, you’d need a very high bandwidth connection and time enough to use it lots, perhaps accomplished at a limited number of “hard line” connection points.

Not that I think shooter situations would be common in an em world. But If you want to make a realistic em shooter, these would be how.

GD Star Rating
a WordPress rating system
Tagged as: ,

Em Econ @ Yale Thursday

The Yale Technology & Ethics study group hosts about one talk a month on various futurist topics. Amazingly, I was their very first speaker when the group started in 2002. And this Thursday I’ll return to talk on the same subject:

The Age of Em: Social Implications of Brain Emulations

4:15-6:15pm, May 22, Yale ISPS, 77 Prospect St (corner of Prospect & Trumbull), Rm A002.

The three most disruptive transitions in history were the introduction of humans, farming, and industry. If another transition lies ahead, a good guess for its source is artificial intelligence in the form of whole brain emulations, or “ems,” sometime in the next century. I attempt a broad synthesis of standard academic consensus, including in business and social science, in order to outline a baseline scenario set modestly far into a post-em-transition world. I consider computer architecture, energy use, cooling infrastructure, mind speeds, body sizes, security strategies, virtual reality conventions, labor market organization, management focus, job training, career paths, wage competition, identity, retirement, life cycles, reproduction, mating, conversation habits, wealth inequality, city sizes, growth rates, coalition politics, governance, law, and war.

My ’02 talk was controversial; Thursday’s talk will likely be well. All are welcome.

Added 28May: Audio, slides.

GD Star Rating
a WordPress rating system
Tagged as: , ,

Factory+Files Future

The difficulty of practical interstellar travel is horrendously underestimated. … Known physics will never deposit living people on Earth-like planets around other stars. (more)

That was Donald Brownlee, who said something similar in our film. It occurs to me that skepticism about cryonics and interstellar travel have similar roots, and that understanding this is useful. So let me explain.

Imagine that one tried to take a rock, say this fossil:

Fossil

and put it somewhere on Earth so that it could be found in a million years. Or that one tried to throw this fossil rock so that it would pass close to a particular distant star in a million years. Few would claim that doing so is impossible. Most would accept that these are possible, even if we require that the rock (plus casing) remain largely unchanged, i.e., retain its shape and maybe even most of its embedded DNA snips.

So skepticism about making people last a long time via cryonics, or about getting people to distant stars, is mainly about how people differ from rocks. People are fragile biological systems than slowly degrade with time, and that can be easily disrupted by environmental disturbances. Which justifies some doubt on if the human body can survive long difficult paths in space-time.

So why am I more hopeful? Because there are (at least) two ways to ensure that a certain kind of object exists at certain destination in space-time. One way is to have an object of that kind exist at a prior point in space-time, and then move it from that prior point to the destination. The other way is to build the desired object at the destination. That is, have a spec file that describes the object, and have a factory at the destination follow that spec file to create the object. One factory can make many objects, factories and files can be lighter and hardier than other objects, and you might even be able to make all the particular factories you need from one smaller hardier general factory. Thus it can be much easier to get one factory+files to a distant destination than to get many desired objects there.

Yes, today we don’t have factories that can make humans from a spec file. But if our society continues to grow in size and abilities, it should be able to do the next best thing: make an android emulation of a human from a spec file. And we should be able to make a spec file from a frozen brain plus a generic spec file.

If so, a frozen brain will serve as a temporary spec file, and we will be able to send many people to distant stars by sending just one hardy factory there, and then transmitting lots of spec files. The ability to encode a person in a spec file will make it far easier to send a person to a wide range of places and times in the universe.

See David Brin’s novel Existence for an elaboration on the throwing rocks with files theme.

GD Star Rating
a WordPress rating system
Tagged as: , ,

Help Me Imagine

For my book on em econ, I want to figure out something unusual about human psychology. It has to do with how creatures with a human psychology would react to a situation that humans have not yet encountered. So I ask for your help, dear readers. I’m going to describe a hypothetical situation, and I want you to imagine that you are in this situation, and then tell me how you’d feel about it. OK, here goes.

Imagine that you live and work in a tight-knit community. Imagine a commune, or a charity or firm where most everyone who works there also mostly socializes with others there. That is, your lovers, spouses, friends, co-workers, tennis partners, etc. are mostly all from the same group of fifty to a few hundred people. For concreteness you might imagine that this community provides maid and janitorial services. Or maybe instead it services and repairs a certain kind of equipment (like cars, computers, or washing machines).

Imagine that this community was very successful about five years ago. So successful in fact that one hundred exact copies of this community were made then and spread around the world. They copied all the same people, work and play roles and relationships, even all the workspaces and homes. Never mind how this was done, it was done. And with everyone’s permission. Each of these hundred copies of the community has a slightly different context in terms of its customer needs or geographic constraints on activities. But assume that these differences are small and minor.

OK, now the key question I want you to consider is your attitude toward the other copies of your group. On one hand, you might want distance. That is, you might want to have nothing to do with those other copies. You don’t want to see or hear about them, and you want everyone else in your group to do likewise. “Na na na, I can’t hear you,” to anyone who mentions them.

On the other hand, you might be eager to maximize your chances to share insights and learn from the other groups. Not only might you want to hear about workplace innovations, you might want to see stats on what happens between the other copies of you and your spouse. For example, you may want to know how many of them are still together, and what their fights have been about.

In fact, when it was cheap you might even go out of your way to synchronize with other groups. By making the groups more similar, you may increase the relevance of their actions for you. So you might try to coordinate changes to work organization, or to who lives with whom. You might even coordinate what movies you see when, or what you eat for dinner each day.

Of course it is possible to be too similar. You might not learn anything additional from an exact copy doing exactly the same things, except maybe that your actions aren’t random. But it also seems possible to be too different, at least for the purpose of learning useful things from other groups.

Notice that in tightly synchronized groups, personal relations would tend to become more like group relations. For example, if just a few copies of you did something crazy like run away, all the copies of your spouse might worry that their partners may soon also do that crazy thing. Or imagine that you stayed at a party late, and your spouse didn’t mind initially. But if your spouse then learned that most other copies of him or her were mad at copies of you for doing this, he or she might be tempted to get mad too. The group of all the copies of you would thus move in the direction of having a group relation with all of the copies of him or her.

Now clearly the scenario where all the other groups ignore each other is more like the world you live in now, a world you are comfortable with. So I ask you to imagine not so much what you now feel comfortable with, but how comfortable people would feel with if they grew up with this as normal. Imagine that people grew up in a culture where it was common to make copies of groups, and for each group to somewhat learn from and synchronize with the other groups.

In this case, just how much learning and synchronizing could people typically be comfortable with? What levels of synchronization would make for the most productive workers? The happiest people? How would this change with the number copies of the group? Or with years since the group copies were made? After all, right after the initial copying the groups would all be very synchronized. Would they immediately try hard to differentiate their group from others, or would they instead try to maintain synchronization for as long as possible?

GD Star Rating
a WordPress rating system
Tagged as: ,

Computing Cost Floor Soon?

Anders Sandberg has posted a nice paper, Monte Carlo model of brain emulation development, wherein he develops a simple statistical model of when brain emulations [= “WBE”] would be feasible, if they will ever be feasible:

The cumulative probability gives 50% chance for WBE (if it ever arrives) before 2059, with
the 25% percentile in 2047 and the 75% percentile in 2074. WBE before 2030 looks very unlikely and only 10% likely before 2040.

My main complaint is that Sandberg assumes a functional form for the cost of computing vs. time that requires this cost to soon fall to an absolute floor, below which it will never fall, relative to the funding ever available for a brain emulation project. His resulting distribution has costs approaching this floor by about 2040:

SandbergTimingModel

As a result, Sandberg finds a big chance (how big he doesn’t say) that brain emulations will never be possible – for eons to follow it will always be cheaper to compute new mind states via floppy proteins in huge messy bio systems born in wombs, than to compute them via artificial devices made in factories.

That seems crazy implausible to me. I can see physical limits to physical parameters, and I can see the rate at which computing costs fall slowing down. But having the costs of artificial computing soon stop falling forever is much harder to see, especially with such costs remaining far higher than the costs of natural bio devices that seem pretty far from optimized. And having the amount of money available to fund a project never grow seems to say that economic growth will halt as well.

Even so, I applaud Sandberg for his efforts so far, and hope that his or others’ successor models will be more economically plausible. It is an important question, worthy of this and more attention.

GD Star Rating
a WordPress rating system
Tagged as: ,

Why Neglect Social Results?

For many decades I’ve heard people argue about the possibility of ems, i.e., brain emulations (also called “uploads”). Many like to talk about whether ems are possible, when they might happen, and if ems would be conscious, or whether they would “be me.”  People also love to read fiction set in worlds where there are ems. Almost twenty years ago I wrote a short article on the social implications of a world of emulations — what that would actually be like. But that didn’t kick-start much interest in the subject – most discussion is still on possibility, timing, consciousness, identity, and story settings.

Over the years I’ve also heard many people argue about the possibility that we live in a computer simulation. Twelve years ago I wrote a short article “How to live in a simulation,” on how you should live your life differently to take this possibility into account. That article also didn’t kick-start much interest in social implications. Today, most discussion of the simulation possibility continues to focus on using it as a setting for fiction, on the chances that it is true, on clues for inferring if it is true, and on what it implies for identity, consciousness, physics, etc. There remains almost no discussion of life strategies conditional on a simulation.

I just now noticed how similar are these situations, a similarity that cries out for explanation. I see three somewhat related candidate explanations:

  1. The sorts of people who most like these topics are techies, who mostly don’t believe that social and human sciences exist, and thus aren’t interested in hearing about  applications of such sciences.
  2. People are mainly interested in these sorts of topics as ways to stretch and stress-test their basic concepts. So only people with a library of grand social concepts are interested in using these topics to stretch and stress-test such concepts. There aren’t many such people.
  3. I personally did a poor job of introducing these topics. Had someone more prestigious or articulate done the job, there might well be much larger conversations now about these topics.

Whatever the explanation, this bodes poorly for interest in my more elaborated book-length discussion of the social implications of ems. However, I will soldier on nonetheless.

GD Star Rating
a WordPress rating system
Tagged as: ,

Dvorsky Em Interview

The always thoughtful George Dvorsky just posted an interview with me on ems at io9. Dvorsky calls my scenario “Dystopian”:

These days, people worry about robots stealing our jobs. But maybe we should be more concerned about massive populations of computerized human brains. … Disturbingly, Hanson predicts that human servants could become status symbols among the rich.

“But to survive, most humans would need assets other than their ability to work,” Hanson told io9. “Like stocks, bonds, real estate, or patents. Humans without such assets may or may not get charity from other humans. Em charity to humans may focus on offering humans a chance to become ems.”

Disturbing, no? Hanson made it clear to me that his analysis is designed to forecast likely outcomes, and not those outcomes we currently desire. …

“In virtual reality, ems need never experience physical pain, hunger, disease, or grit,” says Hanson. …

Given these near-Dystopian scenarios, I asked Hanson what would motivate ems and why they would choose to cooperate with traditional power structures. …

As if a lot of this isn’t already grim enough, ems will have to deal with a host of other concerns. For example, a stolen copy of an em mind could be tortured to extract secrets, or enslaved to compete with the original. … Regardless, the future that Hanson describes is a far cry from what many of us are hoping for. Living as a stream of 1’s and 0’s may have its benefits — but a digital utopia has never seemed more elusive.

I wish Dvorsky would made it clearer what parts of this scenario he most dislikes. Then we could talk more productively about how to make it better. Perhaps I should have emphasized that with investments doubling monthly, humans might quickly turn even a small initial investment into a king’s ransom.

GD Star Rating
a WordPress rating system
Tagged as:

Me Talking Thrice

  1. This Thursday March 6 at 4pm I speak at Duke University in the Philosophy, Politics, & Economics (PPE) Seminar, in room 330 Gross (room TBD), on Shall We Vote On Values But Bet On Beliefs? (slides)
  2. This last Sunday I talked to the DC Philosophy Cafe on Em Econ (audio).
  3. Last week I did another interview with Adam Ford, on Futurism (video).
GD Star Rating
a WordPress rating system
Tagged as: , , ,