The Dementia Diaries

Artificial Intelligence

As subscribers to the ONSN know, I have carefully curated this newsletter to be address readers who are witty, urbane, well-spoken, logical, empirical, cautious, deliberate – in short, what passes for “intelligent.” As an aside, if you try to subscribe and find you are unable for whatever reason to do so, well . . .

Today I want to talk about intelligence. As we discussed a few weeks ago, by standardized measures, exactly one-half of the population is “less intelligent” than the other half. This is based on a scale that winds up as a score next to your picture on your college application. As a psychologist, I have had a passing interest in this scale for some years now. I think I am now on firm ground by arguing it is all hooey, to use a technical term. A predicate of any “measure” of intelligence is that it can be evaluated based on knowledge of a set of verifiable facts, such that knowing 2+2=4 is a sign of intelligence, while thinking 2+2=7 is a deficiency. The whole activity of measuring intelligence assumes that there are a set of common “truths” to which we can all subscribe.

I don’t think that’s how real intelligence works. In support of my position, let me introduce you to my dog. This is Daphne:

Daphne has a special kind of genius – she has everyone convinced that she is the stupidest dog alive. Dumb as mud. And by that simple misdirection, Daphne is never held to blame. Tore up my favorite T-shirt? Still a pup. Peed on the couch? We should have let her out sooner. Stole food from the counter? What do you expect? Daphne is do-dah-do-dah-doing her way through a life of treats, belly rubs, and general forgiveness. This animal is by far one of the cleverest of her species, right up there with labrador retrievers. (Clearly, Kristi Noem has figured this out. She shot her dog when it dawned on her that he was smarter than she.)

In fact, all intelligence is artificial. So-called AI, in the tech world, is produced by scraping vast quantities of information from the massive storehouses of knowledge that constitute the World Wide Web. This activity has been uniquely designed by humans who figured out how to create artificial intelligence by – across the course of their own lives – scraping information from everything in their experience and employing that material to create something new. In sum, your intelligence is a measure of your ability to attend, retain, cross-connect and use your experience to generate solutions.

How you do that is going to depend mightily on your environment, your genetics, and your social circumstances, all of which interact. The guy on Wall Street who has the skills to fleece the public and build a vast fortune – a virtual tiger – would probably shortly perish in a real jungle, where actual predators stroll around unmolested. Intelligence is contextual.

Fifty years ago, or so, I embarked on a career that managed to explore most of the dark corners of the criminal justice system in this country. I began as one does, full of righteous Kool-Aid, ready to slay the dragon of Crime. Along the way, I became interested in why some people turn to violence, thievery, extortion, fraud – the epigenesis of criminality, as it were. As you might expect, it is a complicated question, further compounded by the matter of defining “crime.” At the end of that 50-year journey, I came away convinced that crime is wildly over-rated – mostly, it’s a means by which people in power control the lives of others, with almost no justification for their methods.

I have a friend who, for as long as I have known him (60 years), has confidently asserted that humans will eventually create our own evolutionary successor. With that in mind, I decided to ask ChatGPT what it thought crime might look like if AI was in charge:

“The existence of crime under AI rule would depend on how the AI system is designed and implemented. If AI were to govern the world with ethical considerations, robust laws, and fairness, it might significantly reduce certain types of crime by ensuring equitable law enforcement, surveillance, and social services. However, AI systems are created by humans, and biases or flaws in their design could either perpetuate or even exacerbate criminal activity or injustices.”

So, there you have it. AI will do just fine, if humans aren’t involved.