smhwpf: (BuffyAnne)
The idea of the technological singularity, where artificial intelligence overtakes human intelligence, leading to runaway technological growth with unknown implications for human society, is well-established, although how likely it is remains controversial.

There are numerous concerns about the implications of increasingly autonomous computers and robot systems with artificial intelligence. A very important one relates to autonomous weapon systems, or killer robots, that not only operate without a physical human pilot/driver, but which use AI algorithms to make their own decisions about who to target, and when. In the short term, there are all sorts of moral and legal concerns - who is to be held responsible if an algorithm kills an innocent person? In the longer term, the potential for killer robots to turn against their makers and take over the world and destroy humanity. Such a risk may be far in the future, but it seems to me far from implausible, once you start building algorithms that 'work', but in ways that human programmers do not fully understand, there must exist a risk that they will develop in ways completely contrary to the intentions of the programmers. The Campaign to Stop Killer Robots campaigns on just this issue. It seems to me that, as AI becomes a reality, something akin to Asimov's Laws of Robotics becomes a no-brainer.

The other big potential danger that is often talked about, and which is the main subject of this post, is that of mass unemployment as robots replace more and more human jobs. This has long been the case for blue-collar manufacturing jobs of course, but now the middle classes are beginning to sit up and take notice, with cases like the recent decision by Japanese insurance company Fukoko Mutual Life to replace 34 employees who assess insurance claims with the IBM Watson AI system. The Nomura Research Institute estimated in 2015 that half of all jobs in Japan could be replaced by robots by 2035.

Up to now, advances in technology have certainly caused significant sectoral employment problems among workers with particular skills that are no longer needed; the tendency of a Capitalist economy has been to shrug its shoulders at the fate of these obsolete workers and leave them to rot on the dole, if they're lucky. Sometimes, where there are more social-democratic oriented governments, there may be some effort at retraining, reskilling, industrial and regional policy, etc., to provide new opportunities to such workers. So far, however, fears that advancing technology would lead to permanent and growing mass unemployment have proved unfounded; new technologies make some occupations redundant or less needed, but create new ones, and expand the production possibility frontier so that the great majority of workers can still be employed one way or another, but producing more and more output. Not that this is unproblematic, for all sorts of social, economic and environmental reasons, but the majority of humanity has not been thrown on the scrapheap, and indeed extreme poverty continues to diminish.

Perhaps, then, fears of economic doom due to AI are misplaced? In fact, I think it may be worse than most people think.

Starting with economic fundamentals, production (in the economy as it has been up to now) requires a combination of labour and capital. (The latter in a broad sense may include land). Labour is paid a wage, capital receives a rate of return, in the form of profits, interest or rent.

But capital, and the owners of capital, needs labour needs the rest of us, tho great majority of us who depend on our labour for our livelihoods*, in two ways: first as a means of production; you need some combination of people, land and machines; but secondly as a market for the goods and services produced by labour.

This is crucial. Capital does not reproduce itself, does not get a rate of return by some intrinsic magical property, but because there is demand for the goods and services capital helps produce. It is true that the rich themselves form an important market, but that is not enough to sustain the great majority of owners of capital. The owners of Starbucks and Macdonalds could not become rich just by selling to the rich. Even, say, landlords can only earn rent if their tenants are able to pay it, which means they need employment (or government transfers).

But if AI becomes sufficiently advances, this could cease to be true. If capital can create more capital without labour input, that is if robots can build robots, that can in turn do all (or almost all) the necessary work, then those who own capital (robots and the technology that drives them) not only no longer need labour for production, they no longer need to mass-produce products and services to be sold to the majority of the population. Their capital can provide them with all the necessities and luxuries they desire, and continue to reproduce itself to greater and greater levels of sophistication. No doubt a small number of very lucky humans would be needed to help maintain things (who themselves would quickly join the ranks of the rich, the robot-owners), but the great majority of the population would become completely surplus to the requirements of the elite.

This is a truly terrifying prospect. Would the rest of humanity even be allowed to survive? Perhaps the elite 1% or so would allow the rest of us to continue to eke out an existence as best we could on whatever portions of the earth they decided they had no use for, and without access to the technologies that allow the elite their luxurious lifestyle. They would certainly want to sequest for themselves all the key natural resources they need to keep this new economy running. They would protect themselves of course not only with high fences but with robot armies. They would probably see a need to 'cull the herd' periodically of the roaming barbarians outside their protected zones, less we threaten their system in some way. I suspect they would quite quickly come to see the rest of us as less than human. Maybe some of them would extend 'charity' to a few of us.

Perhaps in such a scenario, a robot rebellion would not be the ultimate fear, but our only hope.

Is there a flaw in my reasoning? There may well be, I hope there is, and please do point it out if so. Or is the point when capital becomes self-reproducing so far in the future that it is not a serious concern for now, especially in the face of other civilization-threatening challenges? Perhaos the Future of Humanity Institute has already analyzed this question, although I did not see anything obviously relevatn on a cursory look at their website.

But if my line of reasoning is correct, then Socialism becomes all the more urgent - that is, the socialization of the means of production, of the technologies that would enable self-contained labour-free production. If capital is all that is needed for production, then we must all own the capital.

The choice will be between fully automated luxury space communism, or the end of humanity as we know it.

*We must also include those of us who do not own capital, but who are unable to work due to unemployment, sickness or disability, or old age. Those of us in this position either depend on our own past labour (savings, pensions), or on a social transfer system that relies on labour income from a large proportion of the population.
smhwpf: (Buffyanne)
From [ profile] katebacross: Given that you have doctorates in both maths and economics and are a former lecturer, how do you feel about the requests from today’s economics students – and Thomas Piketty - to teach economics from a broader subject base, and reduce its reliance on “simplistic mathematical models”?

Oh God, absolutely!

Certainly, Economics can attract many mathematicians because they can very easily be good at it, and publish lots of papers and stuff, but I think that the effect for me of being a mathematician coming into economics is that I see that a lot of the work being done is very pretty mathematics, but not necessarily any actual use to man or beast in terms of understanding how economies work.

The problem is not necessarily maths per se. Right now, the sort of economics that Piketty and others criticise, namely mainstream Neoclassical economics, is certainly incredibly locked into the use of complex mathematical models, but it is not only Neoliberal economics that makes extensive use of maths. Marx's Das Kapital is full of mathematics. But Marx starts with a very different set of assumptions and uses a different set of mathematical tools - based on the assumptions he makes and the type of analysis he is trying to conduct, and thus reaches a very different set of conclusions.

Economic analysis is inevitably going to use mathematics, as it is attempting to deal with quantifiable aspects of human society - how much is produced, consumed, by whom, who gets the benefits. It is a matter of what maths you use, with what starting assumptions.

Neoclassical economics, I believe, is designed to reach a set of conclusions favourable to unrestrained, or as near as possible to unrestrained, Capitalism. Not that its inventors and practitioners are necessarily wicked or dishonest, but that it has gained traction and dominance within academic economics because it favours those with money and power.

tl;dr )

Basically, as I see it, the flaws in the assumptions of Neoclassical economics means that it not only fails to describe how individuals work as economic agents, it fails to describe how economies and economic systems work.

The maths can be very neat and pretty. But it is a very particular type of maths. It is based on a kind of Newtonian world or smooth, continuous, infinitely differentiable functions (of a particularly tractable type), which allow for stable equilibrium solutions. But this is not what the world is like, and the discontinuities, uncertainties, and jagged edges that constitute human life are not mere minor disturbances in the otherwise smooth, predictable fabric, they are the fabric.

Personally, the sort of maths I have most fun with in economics is Game Theory. Some of my most successful economics classes were when I got the students playing Prisoner's Dilemma. But I'm less and less sure how much use this sort of thing really is either, except as maybe a loose metaphor.

But at least that sort of thing actually gets people thinking about human behaviour and relationships, and their implications for economic life.

And maybe this is how I think Economics needs to go, to recognizing itself as a social science; one whose starting point is human interactions within a social system. By all means use maths, but before you know what maths to use, you need to have an initial understanding of the social structures in which economic relations take place, and the ways in which people behave and repsond within those structures. What mainstream Economics does is to assume that all those questions can be brushed aside by imagining a society populated by "Utility Robots", and then in response to the obvious protests say "But it's just a model! It doesn't need to be true, it just needs to predict! On average! You just don't understand our complex mathematics!" and so on.

I think this sort of view is gradually becoming more mainstream? But academic paradigms are slow to shift, especially when Economists are rational publication-maximisers. I suspect that there are a lot of economists who know that Neoclassical economics and the sorts of mathematical models they use are bullshit, but continue to use them because they know how to work with those models and because that's what gets published in the top journals. I do know that at my old place, University of the West of England, a fair few of my former colleagues are really trying to broaden the curriculum to include and even emphasize other perspectives. Strength to their arm.

This article lacks references. I will rectify this and try to find some suitable articles to back up some of my ramblings, but right now it is already late, and I thought it more important to get the thoughts down first. So references will follow, hopefully this Advent, but I do not promise when.
smhwpf: (Homework)
The mainstream media is well known to be rubbish when reporting science. (Ben Goldacre provides a running commentary on that). They are equally rubbish when it comes to financial matters. OK, you can get some pretty clued up commentators. But I'm thinking in particular at the moment about reporting of budget cuts. (A topical matter what with all these austerity packages and all).

An article might say, for example, "The government will cut the budget by €10 billion by 2015" or "make savings of €10 billion by 2015" or something.

Sounds clear enough. The problem is, this can mean one of eight different things. And they never, ever say which.

The first confusion is whether the figure refers to the cut in the annual budget, or cumulative cuts. The first is the simplest case: say the article is in 2011 and the budget is €100 billion, this would mean that in 2015 the budget will be €90 billion. But if the figure is cumulative, it means that €10 billion less will be spent over the whole period 2012-15. So, for example, the budget could go down by €1 billion a year, saving €1 billion in 2012, 2 in 2013, etc. for a total of €10 billion.

Then, they never say whether the cuts are relative to current spending levels, or future plans. So, in fact the big cuts talked about in the US budget are relative to previous planned spending, which in some areas (notably Defense) had been planned to increase. So, if they say there will be "€10 billion cuts by 2015", but it means cumulative and the previous plan was for spending to go up by €1 billion every year, then what that actually means is that the budget will stay constant. Thus, €10 billion less in total is spent over the four years than had the original plan been stuck to. They keep saying "$450 billion cuts to US defense spending over 10 years", but it's cumulative and relative to plans, so in fact the budget is not going to be cut at all. (Though the extra budget for war spending will be as Iraq has ended and Afghanistan is winding down).

The third point is whether the cuts are in nominal or real terms. That is, whether they refer to the actual raw numbers, or whether they are accounting for inflation. As inflation is generally positive, a €10 billion cut in real terms is less than a €10 billion cut in nominal terms. For example, the planned cuts in departmental budgets under the UK's plans are expressed in real terms. But again, this is very rarely specified in media reports.

So with these three variables, that is potentially eight different things that could be meant by the same statement, all with very different implications for what is actually happening. (Perhaps less in practice, as some combinations would be a rather unlikely way for figures to be expressed). So if you want to know which, you generally can't rely on the media reports to tell you, but must go to the budget documents. The problem is, those tend to be LONG. (And for my purposes, of trying to find out about military spending plans around the world, very often in a different language).

Media inaccurate in reporting technical matters shocker.


smhwpf: (Default)

August 2017

  1234 5
13141516 171819
20212223 242526


RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 22nd, 2017 01:15 pm
Powered by Dreamwidth Studios