What is Watson? Strong AI and Higher Education

Watson, for those of you who have spend the last week under a rock, is an IBM computer which soundly trounced two long standing human champions in the US quiz show Jeopardy last week. The Watson story is a good hook for me to jump out of mundane Higher Education policy and get back to some bright splangly futurism!

Watson is one of the newest incarnations of a weak AI - an artificial intelligence with limited scope and capacity, below human levels. These are increasingly abundant things. They beat us at chess, decide on our creditworthiness, keep our cars going, or even drive them for us, trade on stock markets and so on. A great many human jobs only needed 'Weak AI' levels of function anyway, and they have simply vanished, or were never created. Our world economy runs on a vast network of invisible switchboard operators, filing clerks and so on, invisible in the machines. There would be billions of them, but for the machines.

Strong AI - Artificial intelligence on a human equivalent level is a different matter. Like Moon holidays and Aircars, Science Fiction promised it to us a half a century ago, and it never came. Moon holidays and Aircars were disbarred by economics and physics - they could be made to work, but never at a useful price. But the same forces, economics and physics, that stole these dreams from us brought Moore's Law. This rule of thumb predicts the doubling of the available processing power, at a given price, every 18 months. That makes strong AI inevitable. You can argue when, but not if.

Strong AI will mean the end of Universities as we know them, but perhaps also their rebirth as we dreamed them. To understand why, we need to unpack the economics of first decade or two of a world with strong AI.

One fine day, in our lifetimes, IBM, or HP, or some tech giant unborn, will unveil a strong AI. It will be able to pass a Turing test, and will do so for our entertainment on Oprah, The Late Late show, or wherever. It will hold it's own at Go, write a technically competent Sonnet and then quickly fade from the news cycle. Kurzweil predicts a date of around 2029, others later (there is a famous bet on it). It's development will have cost it's company around US$100 million in today's money, that being about as big a budget as a high risk project can justify and sustain. Most of that cost will have been payroll, the hardware will only be a fraction of that, perhaps US$10m (the Watson hardware will cost you about US$3m).

Let's assume that strong AI is about equivalent to a new graduate. It will have relative strengths and weaknesses compared to us 'meatbags' of course. It can read the manual quickly, but might not be so good at charming potential clients. But it probably won't sleep, take holidays, lunchbreaks, or gossip by the water cooler either, so in terms of raw hours it should be about 10 times as effective as a human. If we take a graduate salary of say, $30,000, and an initial cost for a strong AI hardware at US$10m, it's not economic. But Moores law will halve the cost of that power every 18 months. So in a decade or so, a strong AI is going to be cost competitive with a graduate hire, with a hardware cost of around US$300,000, equivalent to the first year wages of ten graduate hires that do the same work.

There many, many assumptions here. I haven't factored in software licensing (Open Source Strong AI anyone?), recruitment and training costs. I've not considered overheads for the humans or AI's, or AI downtime (will AI's need to spend 8 hours powered off a day sorting out our memories as we do?).  Nor have I considered out year salaries beyond year 1. It's all order of magnitude guesses, but with exponential growth in available power, an order of magnitude error makes only 5 years difference. I'm dancing past an enormous debate on whether Moore's Law will hold or not, and taking the probable outcome it that it will.  Early in the second decade after you see a strong AI interviewed on the telly, it will be a cheaper alternative to hiring human graduates.

Now, a human graduate takes 4 years to train (on average, assuming a short MSc after a 3 year degree), and another year before that to get college entry exams sorted out. That only leaves five years after graduation to earn back the cost of your University education, if a strong AI exists before you start. Even allowing a few years slack to uptake of AI's, unless you are already in college when you see that strong AI launched on the news, don't bother going. If you planned to do so to help you get a job, it's too late. Even if you get a job,  you won't make your degree investment back in time before you are replaced. At best you'll spend a couple of years as a human buddy to an AI, until the HR AI figures out that your presence is no longer reducing the error rate, and you are gone. They'll hire a human to fire you. There's a sensitivity subroutine. They're nice like that.

You can still go to college, but go to have a good time. Study Fine Art, or Ancient Persian. Whatever interests and stimulates you. Do Social Work, or Teaching - people centred jobs will be the last to go. Chase your dreams. Learn to Paint, or dance. Meet people. Make friends. Study comparative literature, and sociology. Forget about Business, or IT, or Law, or any of the bankable professions of the olden days. You can't compete.

Our Universities long and often stormy relationship with practicality will be at an end. No longer will they need to bow before Mammon, and produce MBA's and degrees in Marketing or computational Finance. They will return to our dream of them, playgrounds of the mind, where we pursue knowledge for the joy of it, for it's own sake, and not for profit.

(The end of our day to day involvement in economic life may, of course, present other difficulties, which remain out of scope for this blog).

The Tragedy of the Commons and The Last Consumer

As I warned you in the last post, I'm still a little off piste and at the edge of scope thinking about the economic context universities will operate in as the century wears on. Bear with me, I'll stop soon!

My eldest daughter, previously mentioned, wishes to be a Mermaid Musketeer when she grows up. "Wouldn't it be nicer to be a Vet" I think, but I don't say it. I remember how many of today's jobs were (and remain) inconceivable to my father's generation. Maybe Mermaid Musketeers will be in high demand in the 2020's. What do I know.

The conventional narrative of technological development has been that with successive leap forward some gadget or other removes another piece of drudgery from the Toils of Mankind. The newly unemployed riot a little, and then find more fulfilling careers as Advertising Executives, Psychoanalysts and Personal Trainers. Since the plough and irrigation gave us the first agricultural surpluses and allowed priestly and bureaucratic castes to emerge, it's been one of the key narratives of history. Thus, we assert, it will always be so, just as the autumn turkey is confident of a good winters food and a fine spring to come. It ain't necessarily so.

Come with me, if you will, to the supermarket. Tesco, Sainsburys, Walmart, wherever. They in a key place in our world, bringing stuff we need from the four corners of the world into one convenient place, beyond the dreams of any dead King. All strive, rightly, to do so as cheaply and efficiently as possible, cutting costs where they can so they can remain profitable, and competitive on price with the other supermarket down the road. Nothing wrong with that.

A year or two ago, the automated tills were a novelty. People were reluctant to use them, but they have become accepted. It seems slower than the human till, but for a small basket on a busy day, great. I'm sure it means that the supermarket can cut the number of staff at peak times, with one staffer monitoring six or eight autotills. Of course, now that RFIDs are dropping in price, pretty soon we'll just have our trolleys autoscanned on the way out, we can swipe our payment card to exit and be off in moments. Much faster, and it'll be a no brainer compared to waiting in a queue. They can cut most of the till staff. It looks like a horrible job, good riddance.

Meanwhile, back in the storeroom, we'll start seeing more and more machines helping out. It's a lot cheaper to run storerooms with robots. Companies like http://www.kivasystems.com/ are starting to put in place systems that are faster and cheaper to run. Stacking lemons is a bit more complex. It's taken a long time for robots to be able to do that kind of work, but if a robot can fold towels, how far away can a commercial shelf stacker be? A long long time ago, when I was doing my PhD, I paid part of my way stacking shelves for Coca Cola. Great workout No brainpower required.

So as the century wears on, smart supermarket operators will put in those systems. Driven by sales data from the till systems, warehouse robots will load and unload the trucks (no more tricky health and safety issues in the warehouse - no humans allowed) and specialist packer robots will keep the shelves stocked, working mainly at night to minimise human interaction. You could, conceivably, have a complete supermarket shop without dealing with or seeing one human. A nice Augmented Reality system with voice recognition can show you where the cheese is, no shuffling about looking for staff.

There will probably still be a couple of staff though. So many laws assume a shop will have a shopkeeper, it will be hard to avoid having a bored looking manager or greeter around. Technicians may come and go to fix the odd thing, but in time a good R2 unit could replace them. The trucks will still legally require drivers, but as time goes on they will be more closely monitored by expert systems and central controls so they have little or no autonomy. Industry will lobby for UAV trucks to be allowed between, say, three and six am. The accident figures will make their case compelling, eventually.

Of course, in the meantime, most of your food supply will just arrive, a shopping list mediated between the expert systems in your supermarket, your fridge and pantry, the health assist system your health insurer mandates (no more ice cream!(, with a final approving nod from your bank that the delivery fits within the budget you approved. The milk just appears in the fridge, unpacked by your housebot. No more late night runs to the cornershop for milk. Indeed, no more cornershop,as the few that survived the death of the newspaper close up.

In this future, who actually works in the supermarket? We have a few drivers and perhaps a half dozen staff per megastore so there is enough to cover 24/7 opening, annual leave and so on with always one person instore. We would imagine teeming head office, but as AI's and expert systems improve, we need less and less there. Tasks are hived off to expert systems or outsourced to some up and coming service provider where brains are cheap. Productivity per worker, as measured, becomes immense. There just aren't that many workers anymore.

The supermarket story sounds trivial as presented, but you can, with a little imagination, infer a similar story in many industries. A large proportion of our jobs are semi skilled, and do not really demand much brainpower. All the unskilled and semiskilled people who, even in the first world, make the service sector hum are going to be in trouble.

Now, your local supermarket is still making money. It's still paying people, just much less, more highly skilled people, and of course larger dividends to the owners. Who, exactly, is shopping in this supermarket, and with what? All those unemployed people? Henry Ford is alleged to have paid his workers over the odds, as he felt anyone working for him should be able to afford the cars they are making. What's happening here is a parody of that. With each reduction in workforce, there are less and less consumers who can actually afford to buy very much. It's like the Tragedy of the Commons. In this classic economic fable, it pays each farmer to graze the commons as heavily as possible, even though, in the long run, it will destroy the grazing and ruin them all. Increasing automation to increase productivity and cuts costs is a sensible, responsible decision for any business. Each time it happens, it reduces the pool of gainfully employed consumers until there are none left. So whose left with money to shop? Only a handful of highly paid core staff, and the shareholders, mainly pension plans for people who'll never be able to afford to retire.

Historically, of course, the displaced labour has migrated to newer and more interesting professions, but as the machines get smarter and smarter, the pool of professions that only humans can do gets smaller and smaller. I've already blogged about Emily Howell, the virtual composer and other examples of Artificial intelligences tackling problems long thought to be human only. It's also worth noting the set of problems faced by a business are not all best solved by a brain designed for staying alive on the savannah. Intelligences not as smart as us, but different, might do just fine. Think of chess as an example. Or sorting post, or telephone switchboard operators. The machines may even do better, since they lack some of the human brains many, many cognitive bugs. They don't have to be as smart as us, they just have to be smart enough. And besides, who says we're that smart?

Science Fiction writers readily paint pictures of utopian post scarcity societies, where humans live in abundance. Roddenberry's Federation is the classic example, or more recently Iain M. Banks' Culture Novels. The question unanswered is how do we get there from here. The technological path is clear, tractable, and generally plausible. There is however no guarantee that our economic model will be able to adapt to it. Changing economic models is a somewhat risky operation.

Human history has, of late, been an extraordinary positive narrative. While the History Channel drones on about the great wars of the 20th century, we as a humans live in unprecedented numbers and affluence. Famine, poverty and war, once the global norm, as seen as failures, problems to be contained and solved, not accepted. Much of this prosperity comes from technological change. But there is no guarantee that this will continue. It's conceivable that our economic model, structured around rationing and scarcity, might bring us to some kind of dead end. Increasingly homogeneous government models, where each country operates in much the same way following agreed international norms, limits the capacity for different countries to respond in different ways and for new approaches to evolve.

I'm not advocating a stop on technological development. That is impossible, and unwise. We still need to move fast forward to bring the levels of comfort we have largely reached in the first world to all, and solve some of the problems we've created along the way. But we need to be agile and pragmatic about how our societies are organised, and start keeping a good close eye on numbers like the Gini coefficient, so that things don't get ugly. We need to be open for other ways of doing business, and mindful of how we can keep our economic models flexible and adaptable. I'm not preaching anarchism or socialism. I suspect the exact 'ism we will need hasn't been quite invented yet.

As for what it means for Universities, it's hard to tell. In the long run (and I'm thinking a century out here, at least), I think there will be big shift away from professional/ vocational training we see a lot of now, where the focus is often on getting a job at the other end. In a world where there is no job at the other end, or at least, nothing you or I would think of as a job (is blogging a real job?) what people will do in Universities might look a lot more like recreational activity to us today.

That seems like a big leap, but look at our world through the eyes of an early graduate of Bologna or Oxford. Our Universities might look pretty easy to them. No memorisation, no hand copying books. And the jobs out the other end? I don't know how many hours a scribe to Emperor Barbarossa worked, but I suspect they worked harder and longer than a 21st century middle management white collar type.

We're a little further along the road than we might think.

The Four Forces: Driving Change to 2100AD

Four great trends will drive change in Tertiary education to 2100. I've introduced them in previous posts, but let's take a minute to line them up:

Demographics: 10 billion, mostly old people. World population will stabilise at around 10 billion people, and they will be increasingly old. An average age of 55 is not unreasonable by 2100. Longer lifespans will bring more people back for second and third dips into tertiary education, or indeed continuous education.  Overall, the sector might be ten times as large as it is today. It is not unreasonable to suppose that a large portion of the population over 18 might be engaged, in some form, in tertiary education.

Economics: The end of scarcity. A continuation of the 20th century trend would bring another tenfold increase in per capita GDP, making the world, on average, as rich as todays richest country (Norway). Only people at the very margins of society will be unable to afford tertiary education. In the first half of the century, vast cohorts in the old 'Third World' will want, and be able to afford, University educations.

Telepresence: The Death of Distance. Increasingly compelling, immersive and reliable telepresent environments will render the idea of bringing people together in one physical space for education or work a quaint anachronism. Teams or classes may come together once or twice a year, for novelties sake, but true telepresence will make geographic distance as old fashioned an idea as posting personal correspondence in physical mail..

Artificial Intelligence: Smarts too cheap to meter. The steady process of Moores law will create machines with processing power to match the human mind relatively early in the century. Distributed processing will allow systems to draw on immense processing power when needed, and present machines as cheaper alternatives to most jobs currently done by humans. User interfaces that can pass a Turing test will make machine staff indistinguishable from humans. Why hire a human receptionist to answer the phone when the phone comes a processor that can do the job, that doesn't need coffee. When a 1000 euro machine is smarter than anyone you can hire, why hire anyone? The consequences for economics and employment is staggering, and managing the transition will be a huge issue from mid century on.

These are simple extrapolations of well established trends, none of which have any major roadblocks in sight. It's difficult to make a compelling case against any of them. As of 2010, these trends have massive inertia behind them -it's difficult to imagine what scale of events could derail them.

That's not to say that nothing else will happen. Few in 1900 would have predicted the ubiquity of Automobiles, air travel or the Internet. But in 1900, the key trends that set the tone of the 20th century - population growth, economic growth and urbanisation, were in motion. Geopolitical events (like the world wars) could not have been forecast in detail, but the logic of industrialisation made it inevitable that great power wars would get bigger, and worse, until they became so destructive and expensive as to be not worth the risk. We could not have predicted the 747, but we could have predicted that economic growth would have made international travel relatively cheap and easy - we might have predicted a super Zeppelin.

Having mapped out these trends, the challenge now is to figure out what the consequences of these trends are for Tertiary Education, not just in isolation, but as these trends interact and interlock. The second challenge is that the future is not path independent. We don't just wake up in 2100, with institutions and people perfectly attuned to it, no more than our institutions  and peopel in 2010 are perfect fits for the world today. As the century plays out, existing institutions will adapt, or maladapt to the changes. Peoples ideas and preconceptions will change, but only in generational slow time. The future is not a destination everyone arrives at once, it's kind of smeared out, as William Gibson said:

"The future is already here. It's just not very evenly distributed"

Machines could never...

"Pigeons outperform humans at the Monty Hall Dilemma", Blogs / Not Exactly Rocket Science Accessed April 6 2010

Apparently, in specific cases, pigeons are smarter than us. Smarter than me. 


Next time you read a gee whizz article about artificial intelligence and scoff, thinking "A Machine could never do that" remember the pigeons. There is some crazy notion that human intelligence is somehow above and beyond anything that can come from the animal or digital kingdoms. It just isn't so. Animal and Digital intelligences are different, optimised for survival in different environments, that's all. It takes us a while to recognise it. I suspect real AI's will be with us for some time, probably in the form of distributed botnets harvesting bank accounts, long before anyone recognises them as such.


Reference: Herbranson, W., & Schroeder, J. (2010). Are birds smarter than mathematicians? Pigeons (Columba livia) perform optimally on a version of the Monty Hall Dilemma. Journal of Comparative Psychology, 124 (1), 1-13 DOI: 10.1037/a0017703

From Darkness, Light

Emily Howell's new album, from Darkness, Light, is not to my taste. It's a bit too ambient and arty. You might say it's a little soulless. But it's not bad for a machine. I can't imagine our first efforts at writing music for whales would be much better. Her purely derivative works, written in the style of existing composers, are very respectible.
Emily Howell is software. She'll get better. The algorithm will improve, react to market trends, and find the musical keys to move our souls.

Science fiction taught us that artificial intelligence would be invented. It would walk onstage one day, refuse to open the pod bay doors, and announce it would be back. That's not how it's working out. Instead it creeps up on us year by year as they machines take over increasingly complex tasks. Few people remember typing pools, switchboard operators and countless other jobs of the past. We think nothing now of using machines as our research assistants, taking dictation, checking for plagiarism. In five year we'll think nothing of usable machine translation, automated essay grading, machines that write our newspapers, check our diagnosis and fight our wars.

When my eldest daughter starts University in 2023, the computer in her hand will have twice the raw processing power as the one in her head. Distributed computing will put power several orders of magnitude beyond that in her reach. The Turing Test, where a computer can pass for a human, may well be passed by the time she graduates. Profitable niche applications, like generating unique third year history essays at €5 each, or gaming financial systems will reach a point where they can pass for human much sooner than that.

The implications of this for how we teach, and what we teach is Universities is profound and largely ignored. We assume the skills we teach in University are magically beyond automation. Since the first water wheel, machines have displaced human effort, and created a surplus of labour found other, better jobs. The plough freed us to be poets, the steelworkers of the 19th century are the knowledge workers of the 21st. But now the island of our cognitive superiority is shrinking as the waters rise exponentially. The 21st century will be a knowledge economy, but it won't be our knowledge.

It will take time for the change to work through. It often takes a generation for an innovation to move from the journals to the shop floor, and institutional change is also often generational too. What is possible often takes a decade or two to become commonplace, but it does eventually. It isn't science fiction

The changes will be radical. First, practical degrees like science and engineering will become less and less economically attractive in mid century. Softer skills, which might be more difficult for machines to replicate will become more economically attractive. Interpersonal disciplines, where humans prefer to deal with humans, like Medicine, or Theatre, will be the last refuge of economically useful degrees. By the end of the century, as we become habituated to dealing with machines, and no longer notice, or care, about the difference, that too will vanish. Through the century, an increasing proportion the jobs in our economy will be unrelated to production of good and services. We have allready made the journey to from having 100% of humans working in food production to only having (in the first world) a handful. Other industries will make that transition too. With nothing left to do, by centuries end our University system will become largely an entertainment system, a place for humans to amuse ourselves and pass the time.

My daughter says she wants to be a Dinosaur when she grows up. I think she might be right.

The Singular Future

There's a useful summary of Ray Kurzweil's predictions on Wikipedia. If you haven't heard of him, there's a TED talk where he presents his ideas. Kurzweil is a little over hyped (There's a movie - The Transcendant Man, and a University, in collaboration with Google and NASA), and widely criticised, but that doesn't make him wrong.

Kurzweil's basic idea is that technological change, in some key areas, is exponential, not linear. Moores law, that processing power per dollar doubles every 18 months is an example of this kind of technological rule of thumb that has held good for many years. Exponential processes, in their late stages, tend to get a little strange, and Kurzweils predictions, inferred from that, rapidly get wierd. That's trouble with exponential change. Humans can't intuitively grasp it. Our minds, evolved for counting bananas and holding grudges, tend to be unable to get a grip on it. The pond might be a quarter full of weed that doubles every day, but we still expect to be able to leave clearing it to next week.

Where Kurweil breaks from many other futurists is the prediction that computers will reach a point where they are smart enough to improve their own design. At this point, their development and intelligence will rapidly accelerate and exceed ours, and the chart of scientific development goes off the scale. Anything is possible at that point, and the machines will send us an eMail to tell us about it, if they have remembered to feed us. According to this picture, few, if any, of the institutions we know of would remain relevant, let along Universities.

Right or wrong in the long term, Kurzweil's predications in the nearer term are a useful cribsheet for the kinds of technological changes Universities must weather in the next century. True immersive virtual worlds and Artificial Intelligances smarter than us are not outrageous predictions for the 21st century, and will have serious implications for Universities as we know them. If you are a young academic, by the time you have fought your way up to a professorship, you'll be at the sharp end in dealing with these things in teaching. Just when you thought you were clever for mastering powerpoint animations and signing up to Twitter, it's going to get a whole lot harder and meaner.

The impact of technology on the structure of the University is a huge topic, and I'll return to it in coming posts where I'll be looking at the implications of specific potential technologies for the University in detail.