Artificial Intelligence: a few musings

2005-06-20
“If it looks like a duck, and walks like a duck, and quacks like a duck, then it’s a duck” (Ancient wisdom)

Unless it’s a model of a duck.

Artificial Intelligence is model building – we want autonomous machines, but the best we can do is build models of autonomous machines.

Eg, an artificial ant – could be made to behave like an ant in many ways, but not as an ant in an anthill, or capable of making more ants.

2015-10-21
It’s probably possible to make an artificial ant that behaves like an ant in anthill. We may even be able to make an artificial ant that can reproduce in some way.

However, “behave like an ant” is not well defined. There are too many behaviours, and some are obviously easier to mimic than others. Nevertheless, it will soon be possible to make an ant-size robot that can navigate like an ant, climb vertical surfaces like an ant, etc.

But it will always be a model of ant, and therefore its behaviour will in some respect will not be antlike, and in other respects will be a bad imitation of ant behaviour. That’s simply the nature of models.

2016-05-15
Intelligence is even less well-defined than “ant behaviour”. We can mimic some intelligent behaviours, eg, sorting, learning correlations, recognising patterns, and so on, which are useful to augment human tasks such as diagnosis of a fault or illness, or finding the data we want. If a task is well enough defined, we can build a machine to do it.

But that’s the problem: “Intelligence” is simply not well enough defined. My notion of it is the ability to apply and adapt existing knowledge and insight to unanticipated problems. Every term in that definition is fuzzy and vague. Anyhow, some people (including me) would argue it’s more of a definition of creativity than intelligence.

Is consciousness part of  “intelligence”? Many people would say it is. A machine that merely solves problems isn’t intelligent, it’s just an algorithm. It’s not enough to know how to do long division, you have to be able to recognise when and why you should do it. And that you are doing it.

“Understanding” is another component of intelligence. Isn’t it? Well, it does have something to do with learning: an intelligent person is one who can make sense of new explanations. “I don’t get it” at one extreme means “I haven’t figured it out yet”, at the other it means “I can’t figure it out”. The latter is a measure of intelligence.

And that’s just three attempts to make sense of “intelligence”. We’re long way from knowing exactly what we mean by “artificial intelligence”. Far enough that we may not even recognise it when we see it.

The arithmetic of repairing a car

Should you trade in an old car or repair it? The accountants tell you that when the cost of a major repair comes in at about the value of the car, it’s time to buy a new one. “It’s costing you more than it’s worth”, they say.

I think that depends on how you look at it. At current prices, replacing an old car with a newer one will cost $200-$400 a month, whether you borrow the money or save for a future purchase in cash. Insurance and licensing will cost the same whether you drive the car or not. Regular maintenance and fuel costs depend on how much you drive. So $200 to $400 over and above those costs buys a month’s worth of car life.

Let’s define a major repair as one that costs at least $400. That’s about the cost of a brake job. So if the brake job lasts longer than a month, you’re money ahead. Buy a rebuilt transmission for, say, $2000. In less than half a year, you’ll be money ahead. And so it goes. Even body work, such as patching holes or a new paint job, can buy you more additional life than the same money spent on a newer car.

I think that repairing (and maintaining) cars to a nearly new level makes huge economic sense. But as a car ages, owners are less and less willing to do that. There are many reasons for that, but to discuss those would make this post far too long.

 

Update 2023: Cost of repair have risen, but so has cost of purchase. So the argument is still valid. FWIW, my current car is 10 years old and drives like new. Repair/maintenance costs have amounted to about two or three months worth of  financing/purchase cost per year.

In Defence of Processed Food

In September 2015, a CBC program on school lunches pointed out that "healthy" choices are difficult because standards were set in the 1940s when the US Army found that it had to reject a large percentage of recruits for being underweight or otherwise malnourished. Modern processed food is too good, it seems, and is making our children obese.
                     
That reminded me of the days when a large part of a family's time was spent "putting up" the preserves for the winter. Fruit was dried, or made into compotes, jams, and jellies. Cabbage was converted into sauerkraut. Vegetables were pickled or boiled nearly to death and put into sealers. As these cooled, the air inside contracted and pulled the lids down into an airtight seal.

One of the major events at Rutzenmoos was the making of sugar syrup and molasses. The women in the household chopped and sliced sugar beets, then cooked them in the big washing kettle, a copper bowl inset into a purpose built stove, which was normally used to boil the washing in a soap and lye solution as part of the weekly washday rituals. It was of course perfectly sterile. The syrup was a golden colour, the molasses were a nice sticky dark brown. I don’t know whether the syrup was further processed to make sugar, I paid little attention to it. I  concentrated on the molasses, whose taste was I can still sense in my oral memory. Wonderful stuff!

Without processed food, we would have starved.  People nowadays have no idea how important processed food is for survival, and even less how much time was spent in processing it. The food industry made processed food cheap and plentiful. Most of them made wholesome food. But as recently as the 1940s and 50s, governments had to pass regulations to prevent food adulteration, or to enforce safer (and more expensive) processing methods on the less scrupulous manufacturers.

In fact, it was our ancestors' discovery of how to store and process food that led to our eventual dominance of the ecosystem. Until people knew how to grow grains and process other food, they could not live in temperate climates where fresh food is seasonal. True, some people have learned how to use technology to live in very inhospitable climates, the Inuit for example; but they survive because, as luck would have it, their prey contains vitamins without which they would die. That, not technology, is what enables the Inuit to live in the Arctic.

The present day reaction against processed food comes largely from people who have no personal memories of how important processed food is for us. The fact that we can get fresh fruits and vegetables year-round has also helped distract people from this insight.

There’s another fact, which perhaps should be better known: Human digestive systems do not do a very good job of digesting fresh foods. Cooking is a kind of pre-digestion. It breaks down cell walls in fruits and vegetables, and degrades the proteins in meats, making both more nutritious for us. Without cooking, we would get a good deal less value from the food we eat. True, cooking also destroys some vitamins, but usually there’s more left over than we would get from the uncooked food. The same is true of calories. Many starchy foods are essentially indigestible until they are cooked.

Processed food has achieved a bad rep. I think it’s undeserved. In fact, it’s because our food is generally so wholesome and nourishing that the fearful among us fasten on any smidgen of evidence that suggests food is not as good as it might be, however trivial the failure is in the larger scheme of things.

Model Railroading thoughts in 1966 (Highball, November 1966)

In 1996-67, I edited Highball, the newsletter of the 6th Division of the Pacific Northwest Region, National Model Railroad Association. Recently I found a piece of it, this editorial I wrote in the November 1966 issue, after the October 1966 fall Meet (which was in Lethbridge that year). Here it is, as written, with a few typos corrected.
----------------------------
We have a rather slim issue this time, a fault I hope to have corrected next time. I was hoping for a report on the Fall Meet in Calgary; I remember asking for one, and almost wrote one myself. I didn't see Dr. Livingstone's mammoth collection of locomotives, and am told I missed something. I believe it. The Heritage Park Tour apparently was worth missing; either that, or the cold froze all charity out of my informant's system, like salt separates from seawater when it freezes. I rather liked the banquet, and not only because I won a prize (it's the first I've won in years). No; but a chat with Ross McIlveen suggested to me that model railroaders as a group are a very nostalgic lot. What we're really trying to do is recapture the first wonder that trains aroused in us. . . We lost our childhood on a railway train, and ever since we've been chasing real, imaginary, and model trains trying to find the one train on which it steamed away never to return. – Now the layout visits: always fascinating. Two of them were definitely train-watchers' layouts, with lots of room to see a really long train (especially on George Oliphant's layout), and just enough complication of loops and scenery to prevent the whole show from being boring. The other three were modelers' layouts, two of them in the John Allen style of total realism; and one an example of English condensed realism, and very typically and competently English, if the impression made by English modelling magazines is any guide. – That last layout made me feel a bit odd in the throat: my first memories of trains as splendidly self-sufficient, wonderful and beautiful machines are set in England, in the pre-British-Rail(ways) days. Mr. Jarrett models LMS, whose marvellous crimson-lake (maroon) I knew only on branchline coaches and short, six-coupled engines. But the lordly green, black and copper of he GWR' s Halls, Manors, Castles and Kings! The shiny, deliciously chocolate and cream carriages. And Oh! the smell, the sudden heat, the shaking earth when the express London-bounded by, the engine breathing hard and light, and the wheels of the train singing on the sixty-foot long bull-headed rails. – I've watched a Hall, a minuscule engine* by our Canadian standards, pass over the level-crossing a block from my home at better than sixty-per, trailing eighteen or twenty sixty-ton carriages behind her. – In 1947, the four great systems, the GWR, the LMS, the LNER, and the SR were combined as British Railways, and for the next ten years the usual color of an engine was black.

The odds that odd things will happen: The Improbability Principle (Hand, 2014)

   

David J. Hand. The Improbability Principle (2014) Say you go to Timmie’s and meet a person you haven’t seen for 50 years. “That’s a one in a million chance!” people say. Which means that in Canada it will happen to about 37 people. If you mean one in a million per year, that’s 37 this year, and about 900 or more during the 20-odd years since I turned old enough to meet someone for the 1st time in 50 years. “One in a million” isn’t such overwhelming odds for or against after all.

     And that’s Hand’s point. We are bad at estimating odds. Very bad. We react with Wows! to many things that we should expect to happen pretty often. Take the meeting a long-ago friend or neighbour: Since we travel much more than we used to, the odds are far better than they used to be that we will come across someone we knew some decades ago, and even better that that we will meet people who know people we know.

     Explaining why the improbable happens much more often than we expect, Hand provides an excellent introduction to probability and statistics. He writes clearly, with occasional glimmers of a pleasantly dry wit. Anyone who gambles should read this book. His discussion of drug testing will make the reader skeptical of pretty well all news about medical breakthroughs. Which reminds me that reporters of dramatic rate increases in something or other almost never give us the actual numbers. Reporting a 100% increase in some rare disease is much more exciting than reporting that 2 more deaths are expected this year in Ontario.

     And it’s much more thrilling to read about a traveller who changes his flight plans, and so isn’t one of the 300-odd who die when the plane crashes into a mountain. We don’t read about the hundreds of thousands of travellers who have changed their flight plans every day, but have never missed being killed in a plane crash.

     One aspect of our innumeracy that Hand doesn't mention is our difficulty imagining large numbers. That's certainly one reason we have trouble estimating odds. John Paulos says some entertaining things about that in his Innumeracy.

     Well done, recommended. ***½

 

 Another review of this book: https://kirkwood40.blogspot.com/2018/01/the-improability-pirnciple-why-we-dont.html

Oddities in print

     Russell Ash & Brian Lake, eds. Bizarre Books (1985) Ash and Lake share a fascination with publishing oddities: titles that suggest oddball subjects, oddball subjects as such, and books that prompt the question What Were They Thinking? This bibliography, with occasional summaries of excerpts, should satisfy anyone with a similar interest, and will provide mild amusement for anyone needing to fill a few idle minutes here and there. Here are few samples:
     The Hookers of Kew (1967; about a family of botanists)
     How to be Plump (1878)
     The Fangs of Suet pudding (1944; a thriller)
     Frog Raising for Pleasure and Profit (1950; includes recipes)
     How to Fill Mental Cavities (1978)
This book itself should of course be included. **½

Scientific ideas we should forget

 

    John Brockman, ed. This Idea Must Die (2015) A compilation of answers to the question, “What scientific idea is ready for retirement?”, posed on https://www.edge.org/ in 2014. Brockman arranges the answers, starting with general ones, then roughly by topic, such as quantum physics, neurology, evolution, etc, and ending with math and statistics. Often, a short sequence of essays reads like a dialogue.

     Most answers are directed at a general audience, which of course includes scientists in other fields. The writers try to explicate how the target concept causes mistakes or worse, what a better understanding would look like, and sometimes what concept should replace the target. A handful read like part of an ongoing dispute between the writer and the other specialists in the field.

     I was pleased to see that many of my objections, puzzlements, and exasperations were confirmed or clarified in these essays. One of these is the wave-particle duality interpretation of some experiments in quantum physics, which I think is a holdover from the days when observations and models made a nice clean distinction between things that rippled through, and things that bumped into, each other. QM equations show that this distinction isn’t much use. It’s nonsense to say that entities are both waves and particles. It would be like arguing that because people sometimes exhibit fear and at other times exhibit joy, that human beings are somehow both fearful and joyful all at once.

     Another of my annoyances is Schrödinger’s Cat. I’m glad to see that Freeman Dyson notes that the wave function isn’t a thing, so it doesn’t collapse. It’s statement of probabilities in some specific context. (Or conversely, it’s a context defined by a distribution of probabilities). An observation measures one of the probable states. At another time, another state will be observed. To argue that somehow all probable states exist at once is like arguing that because Jack is sometimes angry and sometimes happy when he goes to a baseball game, that therefore Jack is both angry and happy until he goes to the game.

     I found some of the best entertainment in the essays dealing with psychology. One writer attacks a concept, another assumes that same concept in order to attack another one. So what’s an non-expert to do?

     However, the overall effect of reading these essays is the somewhat depressing reminder that we all hold erroneous or misunderstood scientific ideas. They appear in news reports and TV punditry hourly, and many of them have very bad effects on public understanding and thereby on public opinion, which in turn limits politicians’ beliefs about what can and should be done.

     Misunderstanding of basic math is nowhere more obvious than in news about statistics. Case in point: This morning, I heard a report on rising rates of STDs in Alberta, a roughly 40% increase overall in the last ten years, with the highest rate increases among the young and the old, and the lowest among the middle aged. Well, without the actual numbers, rate increases are pretty well meaningless. An increase of, say, from 10 to 20 per 10,000 young would be a 100% increase, while from from 100 to 150 per 10,000 middle-aged people would be only a 50% increase. 50% sounds a lot better than 100%, right? But in this example, 50% is worse, since 50 extra cases will cost five times as much as 10 extra cases.

     The final essay, by Paul Saffo, reminds the reader that the more we know, the more unknowns we encounter. Saffo refers to Teilhard de Chardin’s noosphere, the sphere of knowledge. As it expands into the unknown, its surface increases, the contact between known and unknown increases. I developed this idea on my own many years ago, when I thought of the known as an expanding circle. 2D instead of 3D, but otherwise the same. Either way, there will never be an end to the questions we can ask. Even better, there will always be more questions to ask than have already been answered. But Socrates said as much 2,500 years ago. History echoes.

     Highly recommended, as is the website. ****