Tag Archives: Future

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
loading...
Tagged as: , , ,

Will Rituals Return?

Many social trends seem to have lasted for centuries. Some of these plausibly result from the high spatial densities, task specialization, and work coordination needed by industry production methods. Other industry-era trends plausibly result from increasing wealth weakening the fear that made us farmers, so that we revert to forager ways.

An especially interesting industry-era trend is the great fall in overt rituals – we industry folks have far fewer overt rituals than did foragers or farmers. From Randall Collins’ Interaction Ritual Chains:

Only around the nineteenth century, when mansions were build with separate entrance corridors, instead of one room connecting to the next) and back stairways for servants, did the fully private peerless introvert become common. … Until the beginning of the nineteenth century where is no distinctive ideology of intellectuals as withdrawn and at odds with the world. … The marketing of cultural products … put a premium on innovativeness, forcing periodic changes in fashion, and concentrating a new level of attention on the distinctive personality of the writer, musician, or artist. … The political ideology of individual freedom – which arose in a movement concerned largely to break into the aristocratic monopoly on power rather than to withdraw from it – was often blended with the ideology of the freelance writer, musician, or artist. … Alienation, rebellion, glorification of the inward, autonomous self, an oppositional self taking dominant society as its foil – this has become part of intellectual discourse. …

The daily and annual rounds of activity in premodern societies were permeated with rituals that we would easily recognize as such by their formality; living in a patrimonial household in a medieval community (not to mention living in a tribal society) would have been something like what our lives would be if Christmas or Thanksgiving happened several times a month, along with many lessor ceremonies that punctuated every day. … Modern life has its points of focused attention and emotional entrainment largely were we choose to make them, and largely in informal rituals, that it takes a sociologist to point out that they are indeed rituals. (pp. 362-368)

We can plausibly attribute our industry-era loss of rituals to many factors. Increasing wealth has given us more spatial privacy. Innovation has become increasingly important, and density and wealth are high enough to support fashion cycles, all of which raise the status of people with unusual behavior. These encourage us to signal our increasing wealth with more product and behavioral variety, instead of with more stuff. With increasing wealth our values have consistently moved away from conformity and tradition and toward self-direction and tolerance. Also, more forager-like egalitarianism has made us less ok with the explicit class distinctions that supported many farmer-era rituals. And our suppression of family clans has also suppressed many related rituals.

These factors seem likely to continue while per-capita wealth continues to increase. In that case overt ritual is likely to continue to decline. But there is no guaranteed that wealth will always increase. If we find ways (as with ems) to increase the population faster than we can increase wealth, wealth per person will fall. And if wealth falls, we may well see a revival of overt ritual.

I can’t think of a historical novel that makes clear not only how common was ritual and conformity in farmer or forager societies, but how well that comforted and satisfied people. Nor can I think of science fiction stories portraying a future full of beloved ritual. Or any stories that show how lonely and disconnected we modern folks often feel because we lack the rituals that gave deep meaning to so many humans before us. We tend to love novels that celebrate the values we hold dear, but that can blind us to seeing how others held different values dear.

Perhaps the closest examples are war stories, where soldiers find comfort in finding distinct roles and statuses that relate them to each other, and where they act out regular intense synchronized actions that lead to their security and protection. But that is usually seen as applying only to the special case of war, rather than to life more generally.

GD Star Rating
loading...
Tagged as: , , , ,

Trustworthy Telepresence

In a recent Ipsos/Reuters poll, which questioned 11,383 people in 24 countries, about half believed that they would be at a disadvantage in earning promotions because of the lack of face-to-face contact. Previous research suggests part-time telecommuters do not communicate less frequently with managers. … After four years of experience, the average male telecommuter will earn about 6.9% less than a non-telecommuter. (more)

Telecommuting requires the use of various types of media to communicate, such as the telephone and email. Emails have a time lag that does not allow for immediate feedback; telephone conversations make it harder to decipher the emotions of the person or team on the phone; and both of these forms of communication do not allow one to see the other person. Typical organization communication patterns are thus altered in telecommuting. For instance, teams using computer-mediated communication with computer conferencing take longer to make group decisions than face-to-face groups. (more)

Decades ago many futurists predicted that many workers would soon telecommute, and empty out cities. Their argument seemed persuasive: workers who work mainly on computers, or who don’t have to move much physical product, seem able to achieve enough coordination to do their jobs via phone, email, and infrequent in-person meetings. And huge cost savings could come from avoiding central city offices, homes near them, and commuting between the two. (For example, five firms might share the same offices, with each firm using them one day per week.)

But it hasn’t remotely happened that way. And the big question is: why?

Some say telecommuters would shirk and not work as much, but it is hard to see that would remain much of a problem with a constant video feed watching them. Bryan Caplan favors a signaling explain, that we show up in person to show our commitment to the firm. But a firm should prefer employees who show devotion via more total work, instead of wasting hours on the road. Yes inefficient signaling equilibria can exist, but firms have many ways to push for this alternate equilibrium.

The standard proximate cause, described in the quote above, is that workers and their bosses get a lot of detailed emotional info via frequent in-person meetings. Such detailed emotional info can help to build stronger feelings of mutual trust and affiliation. But the key question is, why are firms willing to pay so much for that? How does it help firm productivity enough to pay for its huge costs?

My guess: frequent detailed emotional info helps political coalitions, even if not firms. Being able to read detailed loyalty signals is central to maintaining political coalitions. The strongest coalitions take over firms and push policies that help them resist their rivals. If a firm part adopted local policies that weakened the abilities of locals to play politics, that part would be taken over by coalitions from other parts of the firm, who would then push for policies that help them. A lack of telecommuting is only one of a long list of examples of inefficient firm policies than can be reasonably be attributed to coalition politics.

Some people hope that very high resolution telepresence could finally give enough detailed emotional info to make telecommuting workable. And that might indeed give enough info to build strong mutual trust and loyalty. But it is hard to make very high resolution telepresence feel natural, and we still far from having enough bandwidth to cheaply send that much info.

Furthermore, by the time we do we may also have powerful robust ways to fake that info. That is, we might have software that takes outgoing video and audio feeds and edits them to remove signs of disloyalty, to make people seem more trusting and trustworthy than they actually are. And if we all know this is possible, we won’t trust what we see in telepresence.

So, for telepresence to actually foster enough loyalty and trust to make telecommuting viable, not only does it need to feel comfortable and natural and give very high bandwidth info, but the process would need to be controlled by some trusted party, who ensures that people aren’t faking their appearances in ways that make it hard to read real feelings. Setting up a system like that would be much more challenging that just distributing something like Skype software.

Of course eventually humans might have chips under their skin to manipulate their sight and sound in real physical meetings. And then they might want ways to assure others aren’t using those. But that is probably much further off. (And of course ems might always “fake” their physical appearance.)

Again, I have hopes, but only weak hopes, for telepresence allowing for mass human telecommuting.

Added 3July: Perhaps I could have been clearer. The individual telecommuter could clearly be at a political disadvantage by not being part of informal gossip and political conversation. He would have fewer useful allies, and they would thus prefer that he or she not telecommute.

GD Star Rating
loading...
Tagged as: , , ,

Auto-Auto Deadline Looms

It is well-known that while electricity led to big gains in factory productivity, few gains were realized until factories were reorganized to take full advantage of the new possibilities which electric motors allowed. Similarly, computers didn’t create big productivity gains in offices until work flow and tasks were reorganized to take full advantage.

Auto autos, i.e., self-driving cars, seem similar: while there could be modest immediate gains from reducing accident rates and lost productive time commuting, the biggest gains should come from reorganizing our cities to match them. Self-driving cars could drive fast close together to increase road throughput, and be shared to eliminate the need for parking. This should allow for larger higher-density cities. For example, four times bigger cities could plausibly be twenty-five percent more productive.

But to achieve most of these gain, we must make new buildings with matching heights and locations. And this requires that self-driving cars make their appearance before we stop making so many new buildings. Let me explain.

Since buildings tend to last for many decades, one of the main reasons that cities have been adding many new buildings is that they have had more people who need buildings in which to live and work. But world population growth is slowing down, and may peak around 2055. It should peak earlier in rich nations, and later in poor nations.

Cities with stable or declining population build a lot fewer buildings; it would take them a lot longer to change city organization to take advantage of self-driving cars. So the main hope for rapidly achieving big gains would be in rapidly growing cities. What we need is for self-driving cars to become available and cheap enough in cities that are still growing fast enough, and which have legal and political support for driving such cars fast close together, so they can achieve high throughput. That is, people need to be sufficiently rewarded for using cars in ways that allow more road throughput. And then economic activity needs to move from old cities to the new more efficient cities.

This actually seems like a pretty challenging goal. China and India are making lots of buildings today, but those buildings are not well-matched to self-driving cars. Self-driving cars aren’t about to explode there, and by the time they are cheap the building boom may be over. Google announced its self-driving car program almost four years ago, and that hasn’t exactly sparked a tidal wave of change. Furthermore, even if self-driving cars arrive soon enough, city-region politics may well not be up to the task of coordinating to encourage such cars to drive fast close together. And national borders, regulation, etc. may not let larger economies be flexible enough to move much activity to the new cities who manage to support auto autos well.

Alas, overall it is hard to be very optimistic here. I have hopes, but only weak hopes.

GD Star Rating
loading...
Tagged as: , , ,

Robot Econ in AER

In the May ’014 American Economic Review, Fernald & Jones mention that having computers and robots replace human labor can dramatically increase growth rates:

Even more speculatively, artificial intelligence and machine learning could allow computers and robots to increasingly replace labor in the production function for goods. Brynjolfsson and McAfee (2012) discuss this possibility. In standard growth models, it is quite easy to show that this can lead to a rising capital share—which we intriguingly already see in many countries since around 1980 (Karabarbounis and Neiman 2013)—and to rising growth rates. In the limit, if capital can replace labor entirely, growth rates could explode, with incomes becoming infinite in finite time.

For example, drawing on Zeira (1998), assume the production function is

GrowthEquation

Suppose that over time, it becomes possible to replace more and more of the labor tasks with capital. In this case, the capital share will rise, and since the growth rate of income per person is 1/(1 − capital share ) × growth rate of A, the long-run growth rate will rise as well.6

GrowthFootnote

Of course the idea isn’t new; but apparently it is now more respectable.

GD Star Rating
loading...
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
loading...
Tagged as: , ,

Jones, Beckstead, & I

Nick Beckstead talked with Garett Jones and I on long run consequences of growth. One point is worth emphasizing: if long run growth matters more than today’s suffering, directly helping those suffering today is unlikely to be the best strategy. From Beckstead’s summary:

What are the long-run consequences of helping people in the developing world, e.g. through donating to GiveDirectly?

If the argument for doing this is that it helps with long-run growth, it’s implausible. It seems very unlikely that donations to GiveDirectly are the best way to speed up economic growth. Improvements in the institutions that hold back innovation would seem more plausible.

Programs like GiveDirectly may have some indirect effects on governance, which could in turn have
effects on long-run growth. For example, people who are suffering less because they are less poor might vote better. We should not assume, in general, that any way of helping people has [predictable] long-run consequences on growth. … [Also,] sending resources from high-growth nations to low-growth nations would be bad for long-term growth. (more)

GD Star Rating
loading...
Tagged as: ,

Sam Wilson Podcast

Sam Wilson and I did a podcast for his series, on near-far, em econ, and related topics.

One topic that came up briefly deserves emphasis: robustness can be very expensive.

Imagine I told you to pack a bag for a trip, but I wouldn’t tell you to where. The wider the set of possibilities you needed to handle, the bigger and more expensive your bag would have to be. You might not need a bag at all if you knew your destination was to stay inside one of the hundred largest airports. But you’d need a big bag if you might go anywhere on the surface of the Earth. You’d need a space-suit if you might go anywhere in the solar system, and if you might go anywhere within the Sun, well we have no bag for that.

Similarly, it sounds nice to say that because the future can be hard to predict, we should seek strategies that are robust to many different futures. But the wider the space of futures one seeks to be robust against, the most expensive that gets. For example, if you insist on being ready for an alien invasion by all possible aliens, we just have no bag for that. The situation is almost as bad if you say we need to give explicit up-front-only instructions to a computer that will overnight become a super-God and take over the world.

Of course if those are the actual situations you face, then you must do your best, and pay any price, even if extinction is your most likely outcome. But you should think carefully about whether these are likely enough bag-packing destinations to make it worth being robust toward them. After all, it can be very expensive to pack a spacesuit for a beach vacation.

(There is a related formal result in learning theory: it is hard to learn anything without some expectations about the kind of world you are learning about.)

GD Star Rating
loading...
Tagged as: ,

Who/What Should Get Votes?

Alex T. asks Should the Future Get a Vote? He dislikes suggestions to give more votes to “civic organizations” who claim to represent future folks, since prediction markets could be more trustworthy:

Through a suitable choice of what is to be traded, prediction markets can be designed to be credibly motivated by a variety of goals including the interests of future generations. … If all we cared about was future GDP, a good rule would be to pass a policy if prediction markets estimate that future GDP will be higher with the policy than without the policy. Of course, we care about more than future GDP; perhaps we also care about environmental quality, risk, inequality, liberty and so forth. What Hanson’s futarchy proposes is to incorporate all these ideas into a weighted measure of welfare. … Note, however, that even this assumes that we know what people in the future will care about. Here then is the final meta-twist. We can also incorporate into our measure of welfare predictions of how future generations will define welfare. (more)

For example, we could implement a 2% discount rate by having official welfare be 2% times welfare this next year plus 98% times welfare however it will be defined a year from now. Applied recursively, this can let future folks keep changing their minds about what they care about, even future discount rates.

We could also give votes to people in the past. While one can’t change the experiences of past folks, one can still satisfy their preferences. If past folks expressed particular preferences regarding future outcomes, those preferences could also be given weight in an overall welfare definition.

We could even give votes to animals. One way is to make some assumptions about what outcomes animals seem to care about, pick ways to measure such outcomes, and then include weights on those measures in the welfare definition. Another way is to assume that eventually we’ll “uplift” such animals so that they can talk to us, and put weights on what those uplifted animals will eventually say about the outcomes their ancestors cared about.

We might even put weights on aliens, or on angels. We might just put a weight on what they say about what they want, if they ever show up to tell us. If they never show up, those weights stay set at zero.

Of course just because we could give votes to future folks, past folks, animals, aliens, and angels doesn’t mean we will ever want to do so.

GD Star Rating
loading...
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
loading...
Tagged as: , ,