IBM Watson and Occupy Wall Street

Watson is the biggest AI celebrity of the 21st century.  I came up with that line when I realised that the PhD students at Bath didn't know who it was, or that they should attend the talk that Dale Lane was going to give us on it this past Friday.  Watson beat two extraordinary human players at Jeopardy (youtube link), live on television.  If you are American, you know how big a deal that was, and if you aren't, then you can watch this video from the eighties in order to realise how much of a cultural event Jeopardy is in the USA.

Dale was nice enough to stay around after his talk, and at some point we started discussing the acceptance of AI by society.   Dale said that the videos of Watson that IBM upload to youtube always get very negative, almost hysterical comments concerning how AI will take over the world.  A lot of this is driven by science fiction.  I've had a web page for years about robot ethics and whether AI more generally is likely to damage society (I don't think so).  But with Dale I started talking about Kurt Vonnegut's first novel, Player Piano.  Player Piano imagines a future right at the moment that the scientists and engineers that have been replacing other people's jobs are finally replacing themselves.  It's a fairly dystopian world, and one of the subplots of the novel involves a guy who has already been "made redundant", as the British say – perhaps allowed to retire early would be the better phrase from the perspective of the authorities in Vonnegut's novel.  Anyway, the character manages to mess up his life and others' by having an affair with his neighbour, and the reader gathers that quite a lot of the society in this dystopia are doing things like that, because they get bored of watching television all day.

Dale said in his talk that the driving vision of Watson had been the Star Trek computer, which you could always ask for help or information.  "Computer" was listening but not interrupting, only responding when addressed.  In the Star Trek universe there are still soldiers, scientists, colonists, artists, traders, bartenders and many other professions.  While the original series talks about money ("credits"), later versions of Star Trek imply that culture has moved beyond money and materialism – anything you want can just be synthesised from pure energy (which is in limitless supply.)  People get to make individual decisions about how to "self actualise", how to create meaning for their lives, but no one needs to work to live (or get medical care).

Growing up I always assumed we were moving towards the Star Trek future, which was after all only 300 years in the future of the late 1960s (so 250 years from now I suppose.)  I was interested in the economic transition – would people retire earlier and earlier until they didn't work at all?  That didn't seem like a good solution since many people enjoy their careers.  What made more sense to me was that as we got better and better at being productive and as our economy kept growing, welfare would pay more and more to people who were unemployed, until finally everyone would find the standard of living on welfare acceptable and working would truly be optional.

This isn't as crazy as it sounds.  It was before welfare reform, and I lived in Chicago.  I knew that public housing could be dangerous, but I also knew that welfare families generally had VCRs, which not all working families had.  Now I know that entertainment becomes a much higher priority when you are bored, but at the time I took that was a sign of at least moderate affluence.  I also knew that suburban kids of white-collar workers would take jobs in the city that people on welfare would not.  For example, jobs in art stores paid only $4/hour, whereas being on welfare was roughly the equivalent of making "almost" $5/hour.  Of course, kids working in art stores wound up getting promoted or moving on to better jobs and those on welfare didn't – that's the welfare trap, but that's not the point of this posting.

The point of this posting is why thirty years later welfare hasn't substantially improved, even though technology and therefore individual productivity has.  I'm not hugely sympathetic to socialist / economic-egalitarian arguments.  I think a lot of people are badly misinformed about the nature of the world.  The twentieth century saw a lot of experiments in communism.  Basically it seems that when people got properly paid for the labour they've already been doing, things went very well for a generation or two. But by the third generation corruption sets in, and a sense of entitlement that led to loss of productivity and therefore economic unsustainability.  Money doesn't grow on trees, and if it just gets printed by the government it isn't worth anything.  Someone has to be producing value, and humans have to be motivated to work.  They are strongly motivated by differentials between each other.  That's what the twentieth century showed.

But I read something in the aftermath of Steve Jobs' death that made me realise there really is a significant problem with employment right now.   Wealth really is increasing while the number of jobs decreases, and both trend are way more profound than I'd realised. [update 2020: actually, the number of jobs wasn't declining, but inequality was steeply rising.]  Well, OK, that's what we expected, especially if you only call it a job if it's in a factory or something.  The problem isn't this, but what is not happening – the wealth or income of people who are not working is not increasing.

Somehow neither Vonnegut nor the Star Trek writers anticipated what's happening now.  And how could they?  After all, we live in a democracy.  How could the USA select a course that wouldn't ultimately benefit the majority of its people?  I think the problem here is very fundamental, the most fundamental problem on the planet right now.  Democracy is the best form of government we know, probably because the possibility of getting voted out is one way to limit corruption.  But corruption can still happen.  The main current problem is that  we have gotten very, very good at getting people vote against their own interests.  Manipulating people's beliefs and desires is a skill we've been developing for at least as long as we've had language.  The challenge is continuous – whenever one solution is discovered soon a counter-strategy is as well.

The problem here may sound like tyranny of the majority, a well-known issue in democracy.  Maybe just like non-smokers, when welfare recipients become more than half of the population everything will get nicer for them.   But actually I think the Occupy Wall Street people are right – opinions on both smoking and taxes were driven from a small minority that were profiting, until the majority of people somehow believed them.  Normally in a democracy minority interests should have influence since minorities matter in close votes.  Something really has to be actively happening to get the majority to ignore the urgent needs of a substantial minority.

So, getting back to Watson, AI and industrial technology are not really the problem.  The problem is political.  Not that politics is bad – politics is just when there's more than one person in the room.  But figuring out how to run a political system is a constant, ongoing puzzle.  Jobs in themselves aren't really a problem either – in a world where everyone really wanted hand-crafted furniture or art as much as in this world they currently want low taxes, then there'd be work for everyone.

Speaking of robots, one thing I've learned from consulting for the European Commission is that Europe has built and sold most of the industrial robots in the world.  However, they aren't in Europe, largely because of the strength of labour unions.  European Labour thinks it is more important to keep jobs than to be efficient, and maybe they are right – I certainly enjoy most aspects of the European standard of living.   Certainly after the original industrial revolution people lost jobs, and probably some of them starved and died.  But now, people in general really are way better off.  We are healthier, we live longer on average, our babies are less likely to die.  That's something political science has learned – one of the best indicators that a state may collapse (that people are so unhappy with their government they may get rid of it) is infant mortality.   Health is a very good indicator of happiness.  So yes, I still think we could have another industrial revolution and eliminate a lot of tedium, and we could all generally be better off after it.

And I'm not saying this sitting from the side of the elite who won't be replaced.  I've been watching the Stanford AI lectures on the Internet and wondering how much longer governments will think they need as many universities as they have now.  And speaking as a cognitive scientist, it's my professional opinion that Watson's intelligence and knowledge discovery isn't that different from human intelligence, but with a much broader access to / memory for existing knowledge.  So scientific discovery might also be automated in my life time. 

So now that we all know we're in this together, let's get back to the challenge.  Honestly, the challenge is not only political, but cultural and philosophical – biological and sociological.  Can we find ways to be actualised – to experience a feeling of self worth – when we know our labour isn't needed?  To be honest, millions of people face this question every day already, either because they are unemployed, retired, or banned from work for some other reason (like their gender.)  What's changing is only that more people need to find answers.   At least, I hope we find answers and not just better and better distractions.

Comments

Joanna Bryson said…
This comment has been removed by the author.
Joanna Bryson said…
Mark Swanborough pointed out that the 4 November Economist had an article quite like mine. Well, similar anyway in that it talks about AI & the industrial revolution. I'm still sceptical that the difference between these industrial revolutions is economic rather than political or demographic, but I could be wrong. But I'm totally certain that the second-to-last paragraph is wrong. If you think AI can't do insight, you don't understand what Watson had to do to solve Jeopardy.
Joanna Bryson said…
For what it's worth, The Economist published my response to their robot article: http://www.economist.com/node/21556884
Anonymous said…
Joanna: excellent summary, and I think right on point. We do mobile security robots, and one of the biggest hurdles we see is the socio-politcal. It is less important that the robot can do the dull, dirty, and dangerous night-shift security job, a key question is displacing workers. But, as you point out we have done this before, there is turbulence during the transition, but things smooth out, with more goods and services available at lower costs to more people. Whether this is a good thing or not I leave to the social scientists :-)
Joanna Bryson said…
Here's a related concept: basic income http://simulacrum.cc/2013/07/10/three-trends-that-push-us-towards-an-unconditional-basic-income/