Trego's Mountain Ear

I’m looking at reports of Artificial Intelligence citing things as fact “that ain’t necessarily so.” The basis for scientific method is to test assertions, that we need to verify our beliefs – testability is the determinant. Time was that bad air was believed to be the cause of malaria – we now know it’s a Plasmodium that is spread by mosquitoes. There are times when common sense just doesn’t go far enough to predict how changing one thing will affect the whole.

I recall the lessons of the Kaibab deer – a spot where reduced predators and domestic grazing competition allowed the deer population to grow to a level where they were starving. The first level interpretation was that reduced predation would result in more deer. True enough. The second level thought was that the greatly increased deer population would exceed the Kaibab’s food production. True enough – and matching Thomas Malthus essay. Still, eighty years after these more obvious results were observed, later researchers were discovering the long-term effects that the huge deer populations had on the area’s Aspen trees. Biological intelligence and natural ignorance can combine very effectively.

There’s the example of anthropic climate change – the basic number can be calculated. A gallon of gasoline is equivalent to about 30,000 calories or 114,000 British Thermal Units. Since we’re using – or to use a more nuanced word, burning – about 100 million barrels of oil annually, those calories pretty well have to be warming the planet. That conclusion is fairly obvious for biological intelligence. It’s probably just as obvious to artificial intelligence. Likewise, it’s fairly obvious that, without humanity and our recent technological progress, we wouldn’t be burning nearly that amount of oil. The problem is figuring out how much it affects things and where. Natural stupidity stops the thinking process with a conclusion that this is horrible and we’re all going to die. We need more analysis to better understand how much threat anthropic global warming is – or, if it might combine with the farmable lands in Canada and Russia to provide huge food surpluses. I don’t have the answer – but I don’t believe Greta Thunberg understands the question.

The strength of an Artificial Intelligence application is that it can review the literature on a topic with amazing speed. The weakness is that – so far, anyway – the application isn’t nearly so good at telling good data from bad. It’s not enough to be the fastest – you also need to be the most accurate – and Artificial Intelligence needs to be protected from natural stupidity.

c