Econ of AI on BHTV

Karl Smith of Modeled Behavior and I did a blogging heads tv show on the economics of artificial intelligence:

It was a pleasure to talk Karl, since he is that rare combination: someone who both takes powerful future technologies seriously, and who understands social science. (Watching it now, I suspect that if you counted minutes you’d find I talked too much – sorry Karl.)

I made an analogy between three ways to grow a nation, and to grow a mind. Growing nations:

  1. Play the usual game of trading with other nations, etc.
  2. Develop good internal support for investment & innovation.
  3. Move all your people to become part of a rich nation.

Growing minds:

  1. Play the usual game of writing code to do more things well.
  2. Develop a super learning algorithm to grow from “scratch.”
  3. Copy an existing human brain, via whole brain emulation.

When possible, I favor approach #3.

I also made the point that while people like to justify having fewer kids in terms giving each kid more help, the factors that seem to influence the choice of zero vs. one kid seem pretty similar to the factors that influence some vs. more kids.  This fits better with the choice really being about more for parents vs. more for the kids. Anyone know of hard data on factors that influence zero vs. one kid relative to some vs. more kids?

GD Star Rating
Tagged as: , , ,
Trackback URL:
  • matthew fuller

    If the ems want to be rich but lack the resources and power, why wouldn’t they combine their resources to self-modify and become more intelligent and powerful?

    Wouldn’t populism be the best option?

  • Robert Koslover

    Thought provoking.
    Oh, and my wife (who was listening in the other room) commented, at one point, “…sounds like a couple of hunter-gatherers whining about how agriculture will make everybody poor.”

    • Carinthium

      To be fair, if hunter-gatherers had the foresight to do so they’d be right- most peasants tilling the land were poorer then their hunter-gatherer ancestors had been.

      Not actually relevant to the topic, but correcting a possible historical misconception.

      • Rob Spear

        Not in terms of possessions.

  • At the BHTV forum I wrote a long comment with lots of annotations linking to previous posts. Rather than repeating it here, I’ll just link.

  • Uploads – “within a century or so”?!

    Coming around to the idea that machine intelligence might well come first?

  • Re: “The artificial bodies and senses are far easier than the artificial minds, so that is the focus of efforts.”

    Probably false, IMO. “Manna” has machine intelligence ordering humans around – because brains came on before bodies. That is certainly how things seem to be playing out at the moment. The machines arrange the human schedules, task management and investment direction, but robots can’t yet match the human body in many areas – and so they get them to take photos, videos, recordings, etc for them with their advanced molecular nanotech bodies to stoke their servers with training data. Brains first, bodies second.

    A brain is just a universal computer – and some fancy software. Bodies have moving parts and are more complicated to make and maintain.

  • In asking what influences decisions to have a first child v. decisions to have an nth child, you seem to assume that decisional factors would be important to the outcome. Actually, demographic features seem to massively outweigh decisional factors in childbearing and family size outcomes.

    Perhaps you’re asking about high-income, educated married couples’ decisions, but even so, these account for a small fraction of children actually produced.

  • Sister Y

    Also – here’s a brand new study (press release two days ago) that I think calls into question some of your thinking on the influence of personal wellbeing, status, and children’s well-being on parents’ decisions to procreate:

    Parents rationalize the economic cost of children by exaggerating their parental joy

    • This study just confirms the well-known fact that people will search for and overplay evidence which confirms that they made the correct decision in the past, while ignoring and downplaying contrary evidence. This is especially true for huge, irrevocable life decisions like having children, choosing a spouse, etc. I don’t see anything in the study which suggests that the choice to have children is different than decisions of similar magnitude.

      This doesn’t tell us hardly anything about what influences people’s child-rearing decisions before they make them.

  • Pingback: К предыдущему посту одна ссылка | Sergey Kurdakov blog()

  • consider

    Can Robin tell us where he got the data for the world economy doubling every million years back in the day? I have a similar background, though ten years younger, in that I have a physics degree and am finishing an econ PhD. But we physics types like data. So where is it?

    And where is the data that the world economy doubled every 1,000 years in the agricultural age?

    I’ve read that economic historians claim that in the 1600s in Western Euroupe, the economy grew maybe at 0.1%. Then in the 1700s with better data, 0.2 to 0.4 percent.

    In the early industrial age, Britain grew at 0.5% to 1.0% from 1800 to 1825 or so, then it picked up to 1.0% by 1830.

    Anyway, I’d really appreciate a source to his world growth in 500,000 BC and 13AD

  • Pingback: News Bits()

  • consider

    I found the source…

    In a 1998 essay, Hanson cites DeLong’s essay, “Estimating World GDP, One Million B.C. to the Present” which in turn cites Michael Kremer’s 1993 paper “Population Growth and Technical Change, One Million B.C. to 1990” in the Quarterly Journal of Economics, Aug pp.681 -716.

    So Hanson agrees with DeLong who agrees with Kremer that world growth grew proportionally to population growth up until 1950.


    First, almost all of Kramer’s papers since 1993 have been in health care and some on education, not that it makes the above paper wrong.

    But we have to assume his population estimates way back when don’t have signficant variance among those attempting to measure and then we have to accept his and Hanson’s assumption that growth was proportional to population growth.

    Also, Hanson is saying growth doubled every million years yet the weak data starts 1 million years ago.

    Does anyone else of a problem with the above?

  • snarles

    What warrants your extreme skepticism to the possibility of war occurring due to disruptive technological change? In several of the scenarios you discussed, it would be *hard* to imagine that politics would not be involved.

  • matthew, I’m sure they’ll try, but it won’t be easy.
    Robert, if anyone was “whining” it must have been Karl. 🙂
    Tim, uploads are machines, and I still think they’ll come first.
    Sister, I just said “factors”, and didn’t specify type.
    snarls, i didn’t say war was impossible; I said it was not inevitable.

  • consider

    (posted at BH)

    But in 2000, Hanson used Hawks, et al. who estimated 10,000 humans in 2 million AD. He quotes the authors of that estimate and argues, “A wide range of genetic evidence suggests to [Hawks et al 00] a population of about ten thousand at about 2 million B.C., and that we “cannot reject exponential growth from” then until 10,000 B.C.” But “we cannot reject exponential growth” from 2 million AD to 10,000BC doesn’t inspire much confidence.

    As for even recent GDP data, one might consider Gregory Clarke’s criticism of Maddison’ pre 1820 figures as “a fiction” or something to that effect.
    (This seems like pretty flimsy growth data).

  • Pingback: Overcoming Bias : Brian Christian BHTV()

  • Andrew

    In terms of the labour explosion and the concepts of rights, humans can still gain a lot by examining today’s world. Right now, the rich are very much in the minority, with much of the world on very low wages. Slavery is still practised in some countries, or at least morally so if not legally so, in Qatar for example.
    But today, poor people aspire to be rich.

    What happens when humans are in the minority and these AI can replicate themselves. Why breed humans when you can have far more *boxes* for the same resource cost?
    Suddenly all humans become mere bourgeoisie, with little economic justification for their role.