Webdiary - Independent, Ethical, Accountable and Transparent
header_02 home about login header_06
header_07
search_bar_left
date_box_left
date_box_right.jpg
search_bar_right
sidebar-top content-top

Artificial intelligence and globalisation

Kenneth Rogoff is Professor of Economics and Public Policy at Harvard University, and was formerly chief economist at the IMF. He has also had published on Webdiary, via Project Syndicate, The Indian tortoise and the Chinese hare, Who's dependent now?, and America's perpetual Christmas.

by Kenneth Rogoff

Today’s conventional wisdom is that the rise of India and China will be the single biggest factor driving global jobs and wages over the twenty-first century. High-wage workers in rich countries can expect to see their competitive advantage steadily eroded by competition from capable and fiercely hard-working competitors in Asia, Latin America, and maybe even some day Africa.

This is a good story, full of human drama and power politics. But I wonder whether, even within the next few decades, another factor will influence our work lives even more: the exponential rise of applications of artificial intelligence.

My portal to the world of artificial intelligence is a narrow one: the more than 500-year-old game of chess. You may not care a whit about chess, long regarded as the ultimate intellectual sport. But the stunning developments coming out of the chess world during the past decade should still command your attention.

Chess has long been the centerpiece of research in artificial intelligence. While in principle, chess is solvable, the game’s computational complexity is almost incomprehensible. It is only a slight exaggeration to say there are more possible moves in a chess game than atoms in a universe.

For most of the twentieth century, programmers were patently unsuccessful in designing chess computers that could compete with the best humans. A human chess master’s ability to intuit, visualise, and prioritise easily prevailed over the brute force approach of computers. The computers gradually improved, but they still seemed far inferior to the top humans. Or so we thought.

Then, in 1997, in what will surely long be remembered as a historical milestone for modern man, IBM’s “Deep Blue” computer stunned the world by defeating the world champion Garry Kasparov. Proud Kasparov, who was perhaps more stunned than anyone, was sure that the IBM team must have cheated. He sarcastically told reporters that he sensed the “the hand of God” guiding his silicon opponent.

But the IBM team had not cheated. Rather, through a combination of ingenious software and massive parallel computing power, they had produced a silicon-based entity capable of such finesse and subtlety, that international chess grandmasters worldwide (including me) were simply amazed. Since 1997, the computers have only gotten better, to the point where computer programmers no longer find beating humans a great challenge.

Only a game, you say? Perhaps, but let me tell you this: when I played professional chess 30 years ago (I once represented the United States in the World Chess Championship cycle), I felt I could tell a lot about someone’s personality by seeing a sampling of their games, even those of an amateur. Until a short while ago, I could certainly distinguish a computer from a human opponent.

Now everything changed like lightning. The machines can now even be set to imitate famous human players – including their flaws – so well that only an expert eye (and sometimes only another computer!) can tell the difference.

More than half a century ago, the godfather of artificial intelligence, Alan Turing, argued that the brain’s function could all be reduced to mathematics and that, someday, a computer would rival human intelligence. He claimed that the ultimate proof of artificial intelligence would be met if a human interrogator were unable to figure out that he was conversing with a computer.

The “Turing test” is the holy grail of artificial intelligence research. Well, for me, a chess game is a conversation of sorts. From my perspective, today’s off-the-shelf computer programs come awfully close to meeting Turing’s test.

Over the course of a small number of games on the Internet, I could not easily tell the difference. True, today’s computers have not evolved to the level of the deranged chess-playing HAL in the filmmaker Stanley Kubrick’s masterpiece “2001: A Space Odyssey,” much less Arnold Schwarzenegger-like droids from the Terminator movies. But the level that computers have reached already is scary enough.

What’s next? I certainly don’t feel safe as an economics professor! I have no doubt that sometime later this century, one will be able to buy pocket professors – perhaps with holographic images – as easily as one can buy a pocket Kasparov chess computer today.

So let’s go back to India and China. Globalisation proceeded at a rapid pace through much of the last century, and at a particularly accelerated rate during its last two decades. Yet the vast body of evidence suggests that technological changes were a much bigger driver in global wage patterns than trade. That is, technology, not trade, was the big story of the twentieth-century economy (of course, the two interact, with trade helping to diffuse and stimulate technology, but this is a matter of semantics.)

Are we so sure that it will be different in this century? Or will artificial intelligence replace the mantra of outsourcing and manufacturing migration? Chess players already know the answer.

Copyright: Project Syndicate, 2006.
www.project-syndicate.org
left
right
spacer

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Brute Strength

Chess playing computers use brute strength rather than intelligence to play the game. While good human players must plan ahead and anticipate their opponent's moves, a chess program will search every possible move up to N moves ahead to arrive at the most advantageous of the options. This is computationally intense and the work required to produce a solution expands geometrically with N.

The earliest chess programs used N=2, to provide a response in reasonable time, but they were no match for human players. As computer hardware got faster, N increased, but it still takes a dedicated IBM supercomputer costing millions of dollars to increase N sufficiently to beat a grand master.

IBM's chess champion can't play GO, because the game cannot be solved with search tree techniques. For my money, (and I know, its only two cents worth :-) it is the ability to adapt that defines intelligence and our computer software has a long way to go before it can do that in anything other than the most trivial of ways.

One practical measure of low-level intelligence in a significant software system would be the ability for the software to install itself on any machine (Mac, X86, Alpha, Sun, etc.) and optimise itself for the environment in which it runs. That sort of computer technology is still a long way off.

Intelligence lies in transactions

Intelligence, artificial or animal, lies in doing, not in being. Unless a brain is functioning, it is simply grey goo.

Rogoff tried to link chess playing computers with the potential for artificial intelligence to change industries and economies. By so doing he limited his, and our, understanding of how social and non-human intelligence is evolving.

Think of the humble outback termite - and the magnificent housing towers it manages to construct. Aligned carefully to minimize heat, with sophisticated air controls: how does a termite, with a brain far smaller than a grain of rice, manage to accomplish such a feat? Entomologists, like EO Wilson, argue that it is through the interactions of the termites' social organisation that such group achievements are generated.

In other words, the group of termites has more intelligence than the individual - a whole lot more. This idea is sometimes known as the wisdom of the flock. Similarly, when the hundreds of thousands of players on stock markets make their plays, a group intelligence emerges. Individual players may be wrong. Indeed, it may be that there has to be people who are wrong in order for a right decision to emerge.

Looked at from this angle it is clear that the Internet and the myriad relations it is allowed has already generated many new forms of group intelligence.

Take that one step further and look at how computers have already been programmed to make decisions - based on both human and environmental inputs. The complexity and sophistication of those functions is already considerable and constantly accelerating.

So, what we are witnessing is the emergence of a new form of computer-enabled and mediated group intelligence. It isn't a simple form of intelligence based in one machine or program, by definition, it can't be. Nevertheless it is real and offers extraordinary possibilities...

I think it would be fair to say that most people on this site come from what could be called the 'old left'. Amongst the core beliefs of that group was the idea that centralised decision-making was more efficient than market driven. Hence all the calls for national plans, government set goals etc.

However we now know that such centralised decision-making carries with it extraordinarily high costs and inefficiencies. Getting “old left” types to admit this is difficult enough.

Getting them to acknowledge that group decision making, based on the interchange of information through market structures is more efficient and intelligent, is likely a bridge too far. However ignoring all of the advances in our understanding of the mind - both individual and social - and all the practical successes and advances of the last three or four decades is also not tenable.

Globalisation is not a theory but, as Bill Clinton noted, a force of nature. It grows out of the technologies and evolving social and economic patterns that those technologies allow. The emerging forms of intelligence allowed by these new structures also cannot be ignored. Otherwise we run the danger of creating two worlds: one rejectionist and dumb, the other smart, adaptable and uncaring.

Horse Feathers

Michael ES, even though I am repeating myself, here goes. I am immersed in the software development industry all day, every day. I work on mission critical projects for companies such as Chase Bank, Iron Mountain, Prudential Insurance, Boeing, Xerox, Kodak. Here in Oz, for Coles Myer & Westpac among others.

Both you and Rogoff are the alchemists of this day, hoping that humming the mantras will turn base metal into gold.

So I repeat it again. The only intelligence is in the heads of the programmer's. Computers are just like hammers, tools to do a job. No intelligence lives there except what a programmer put there. Interoperability, neural nets, artificial intelligence  etc are silly ideas looking for silly people to part from their money.

The key to the vibrancy of the Industrial Age and beyond has been proprietary systems doing things that no one else can do. Today's ideas of central repositories and a "one size fits all" intelligent global solution to our challenges is a death knell. The sooner we stop listening to Rogoff and the other snake-oil boosters the better off we will be.

I am currently involved in client project where an intelligent system had been developed by another group to short-circuit the long lead time on getting new products to market. The system is so complicated that I have quoted 3 months to document the current rule base before any further development can be done.

The people who will be required to work this system need a totally new skill set including better education. This in turn will mean hiring more expensive employees. In any case, when we finally walk out the door, the client will have only one person who will fully understand what they are working with. This is a mission-critical failure point. However, there is no artificial intelligence available that can solve the client's need for smarter, dearer employees.

In any case, you cannot hand over your whole business investment and strategy to a machine, no matter how intelligent. People still need to make the decisions. People rule, machines obey anything else is perverse.

It's the system

Hey Roger, au contraire, I know for a fact that computers do things without humans being aware of it. Surely you must be aware of that.

As you are in the software industry you will know that any help provided by those using your software know better than you. Programmers clearly have little control once a system is up and running.

After all I have been told so many times when trying to resolve confusion (particularly with Centrelink) that "the system did it". What more evidence do we need?

Now who knows better than they? If it had only been Centrelink I may have agreed with you but I've been told this by many people from all ranges of organisations.

The "system" is clearly always waiting for people like myself to produce a random response or result.

And I must admit, having been in charge of a help desk some time back, that our list of explanations had at the top "The system did it, sorry". Good excuse that one as then the help desk person can sympathise with the user about what the system does by itself.

Please don't destroy Excuse number 1 Roger, what will help desk people do?

How Silly Of Me

Yes Ross, just as in Toyland, the machines do come alive when no one is looking.

I've not seen it myself but I have heard reports such as the ones that you quote.

It really is a worrying thing, but currently there are about 10 million of us programmers working on the problem. I have every confidence that we will eventually suss it out.

Of course, there is always the ON/OFF button. Computers hate it when you do that to them.

Rogoff comes at this topic

Rogoff comes at this topic from, as he notes, a narrow portal - the extraordinary developments we have seen in the ability of computers to play sophisticated and adaptive chess. Within the clearly defined rules of that game, some computers now routinely outperform humans.

From that he draws something of an instinctive feel for the possible implications of AI developments. He extrapolates to the rather broad, but I think valid, argument that technological change will be more important in 21st century geo-politics than the ongoing rise of China and India. Of course technological change will be the primary driver of geo-political change in this century - as it has been in at least the last three. In fact, technological change - and not spurious and largely irrelevant political and social theories - has been the dominant change driver for a very long time.

Roger Fedyk and Michael Coleman then throw in their two cents worth, opining that computers are just dumb tools - no more intelligent than a hammer.

The crucial development that is now transforming computer networks is the ability of many humans to interact with the whole network. The network as a unit - or, if you like, as a community - then acquires skills and capabilities. It is as if a million hands were holding that one hammer handle.

The crucial development that will transform computer networks in the next decade is the ability to add in vast arrays of sensing devices - devices that can take inputs - not just from deliberate human input - but from the environment as a whole.

We will delegate decision making power to these networks - because it is convenient for us to do so. Current examples are the control systems in aircraft and those controlling power sources in hybrid cars.

Link together these individual decision-making nodes and the network begins to evolve into a decision-taking mechanism. We have already seen that in equity and derivatives trading - where 'program driven' buy and sell orders can make - and break - market trends.

Are these computer networks actually evolving into individual computers with extraordinary powers? Is the network the computer?

The question is far more than just one individual human sitting in front of one machine - controlling all its inputs. Then it is just a dumb tool. But that stage of development ended back in the '70s when the internet began its long evolution.

We are far past that now. Rogoff's insight is valid, if poorly developed and expressed.

Well, well, well

Hello Michael ES Long time no hear.

Of course computers are just tools, like a hammer. I say that as an Electrical Engineer since 1966 and as a programmer since 1976 and as the head of a computer programming consultancy. If I thought, computers were smart I would sack my programmers tomorrow and save a bundle. 

Why am I telling you this anyway? Don't you already know that the intelligence is in the program, end of story. The whole SAP-driven, internet-based, reengineering and data-mining endeavour that is the mantra of big companies is still a unique solution for one company at a time. The overreaching by Rogoff is bogus.

Your nifty little example of a decision-taking mechanism in the equities and derivatives trading markets is child’s-play programming for a journeyman programmer. Somebody (a statistician cum actuary cum bean-counter) sets up the decision table values and a simple program executes the trades. I note, however, that you did slip in a “program driven” as if we would not notice. In fact, if the sloppy and dangerous state of computer trading is an example of artificial intelligence then strap me to a rocket and send me to Mars where I will get a more intelligent conversation with the rocks.

Limited thinking limits understanding

Ah Roger, I once thought you were one of the few on this site with any real intelligence and ability to communicate. You disappoint me.

"The intelligence is in the program, end of story." What archaic 19th century nonsense. As we are learning from neurobiology, intelligence is in the action, or transaction, not resident in any physical or static structure.

I think, therefore I am. Without thought we only have inert grey matter, or meaningless code.

The intelligence lies in the process.

Furthermore, you failed to even register the two critical elements I mentioned - the network effect and sensor inputs.

I feel like I am tapping on a peculiarly muddy and obscure screen. Is there any intelligent life left in this Antipodean sinkhole? Is thought possible amongst such reactionary bigotry?

Hamish: listen guys, I know that Mercury has recently gone into retrograde, but can we all just be nice? 

Very Superficial

I'm with Roger on this one. I've worked with computers for thirty years and AI is one of my pet technologies. I've been actively involved in programming neural network systems for factory operations and genetic search algorithms for complex schedule optimisations. Simulated or superficial intelligence is a better description of this family of programming techniques, IMO.

Most computer users are aware of just how dumb the machine is. A computer will execute a flawed program without any awareness that it is producing nonsense results. It gets no positive reinforcement from executing a perfectly designed and implemented program. It always does precisely what the program instructions command with neither pride nor remorse. It cannot spontaneously adapt to inputs that the programmer did not anticipate.

We humans are pretty good tool makers and computers are most versatile tools, but a computer has no more intelligence than a hammer.

Not So Clever

Kenneth Rogoff is having sport with us. The little game of subterfuge that he is playing is to confuse us as to whether intelligence is remembering a lot of things or drawing inferences from abstract ideas and thoughts.

Computers are computers are computers. They are tools, no more, no less.

No matter how you paint it, a computer is basically an ALU (Arithmetic Logic Unit) coupled to lots of storage and a few other gizmos (please, no discussion on all the I/O controllers, FPUs etc). The storage has two main purposes. One to hold the program that controls the ALU and two to hold all the information that the ALU plus gizmos will churn. If there is any artificial intelligence, it is in the heads of the programmers who created the program.

When a computer can paint something as exquisite as the Mona Lisa or grow a vegetable plot or comfort a distressed baby come back and let me know. Anything else is the product of minds that seem to have little else to do except produce intellectual fluff.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
© 2005-2011, Webdiary Pty Ltd
Disclaimer: This site is home to many debates, and the views expressed on this site are not necessarily those of the site editors.
Contributors submit comments on their own responsibility: if you believe that a comment is incorrect or offensive in any way,
please submit a comment to that effect and we will make corrections or deletions as necessary.
Margo Kingston Photo © Elaine Campaner

Recent Comments

David Roffey: {whimper} in Not with a bang ... 13 weeks 16 hours ago
Jenny Hume: So long mate in Not with a bang ... 13 weeks 1 day ago
Fiona Reynolds: Reds (under beds?) in Not with a bang ... 13 weeks 3 days ago
Justin Obodie: Why not, with a bang? in Not with a bang ... 13 weeks 3 days ago
Fiona Reynolds: Dear Albatross in Not with a bang ... 13 weeks 3 days ago
Michael Talbot-Wilson: Good luck in Not with a bang ... 13 weeks 3 days ago
Fiona Reynolds: Goodnight and good luck in Not with a bang ... 13 weeks 4 days ago
Margo Kingston: bye, babe in Not with a bang ... 14 weeks 1 day ago