Guy Reams (00:01.198)
Today is day 221, AI turkeys and David Hume.
So in my early days of college, I had a professor with this unusual mission. She was attempting to prove to the entire class that eating turkey for Thanksgiving was wrong. She completely failed in her mission, but she had this pet turkey and her aim in her class was to convince us that for the last 200 years, our minds had been shaped by indoctrination.
At first, I thought this was crazy and ridiculous, but then I realized her method. She knew that she would not convince people not to eat turkey for Thanksgiving, but she wanted to use the powers of inductive reasoning and other misconcept and other methods to convince us of our error. And then.
She did this very slowly, and you didn't really know what she was trying to do. She took you through all these different philosophers and walked you through their reasoning and their guidelines. And at the very end, suddenly she brought on all this evidence that the mass incarceration and killing of turkeys was wrong for all of the reasons that we'd been convinced prior through the logic that we had.
In fact, at the very end, this was like the day before Thanksgiving, she brought her pet turkey in and the pet turkey walked up and down the aisles strutting back and forth very proudly and following commands just like a circus dog would or a dog would, right? So she convinced us that this turkey was actually smart. So the idea was that we would not condone the sensual slaughter of millions of dogs.
Guy Reams (01:58.222)
on a special day, because we believe the dogs are intelligent. So why would we do so with Turkey? So this was her her her method. Now, this was, you know, laughed at for the most part. But I have to agree that her meticulous instruction of walking through this logical process really convinced me or really had an impact on me because I remembered everything that she said.
She first introduced this category based induction problem that was part of David Hume's writings. David Hume was a Scottish philosopher in the 1700s. And he was quite radical for his time. He probably became one of the most influential figures in Western philosophy because he questioned the reliance that humans use for inductive reasoning. The idea that
you can derive general principles from specific observations. Of course, this challenged both religion, and it also challenged science at the time. And since that time, scientists and psychologists have been trying to deal with David Hume's challenge. And there's been lots of different ways people have responded to David Hume's challenge. But...
He did make a pretty compelling argument that there's no rational basis for making inferences from a phenomenon from just a few observed details. My college professor illustrated this point through the perspective of a turkey. Every morning the turkey wakes up to a high protein, high fat meal, which is actually really tasteful and plentiful. The turkey's life is one of ease. It wakes up every morning in comfort and safety. Every sunrise, the turkey experiences the same
relatively safe day with lots of food. And so the turkey grows to expect that every day will be the same. Of course, this is fatally flawed because the fourth Thursday of every November, the turkey finds out that not every day is the same. So this is an exaggerated example of the core issue with inductive reasoning. Not every day will be the same. And this concept, although silly,
Guy Reams (04:18.317)
applies broadly to all that we take for granted. We often make many assumptions based on past experience growing comfortable with what Hume identified as an inherently flawed method of reasoning. This has been the state of psychology for centuries. People try to deal with the fact that humans make inductive reasoning all the time, that humans are able to play in the gray area, that we're okay with making flawed judgments as long as those flawed judgments are generally good.
close enough or have reasonable accuracy. Knowing full well it's flawed, but yet able to make category based assumptions. You know, we look at a dog and we assume the dog because it has certain behaviors while the same behaviors as other dogs that are in the same category. This can work both good and bad for us. Sometimes we'll look at people and assume they will act the same way as they are in other categories of people, which can be considered a bad behavior of humans. But,
The point is that we are able to automatically make assumptions about properties and behaviors of one thing in a category, of other things that we put in that same category. Now that sounds rather simple because that's the way our brains are wired, but that is not actually that simple in general. So there's actually been a collision this last few years between...
what we call psychology and people trying to convert that human psychology into mathematical principles and computer science. And that collision is what we're now calling artificial intelligence. And some layman this as a potential replacement for humans or as a challenge to our dominance in the realm of imperfect cognition, because humans have always been the only creatures that were able to do this. And now it seems that we have developed
software that can make the same kind of fuzzy logic or the same kind of jump to inductive reasoning that we have always been able to do very effortlessly. So that has caused us to marvel a little bit, like are we going to see the replacement of humans? Well, I'm assuming that imperfect cognition, we will probably develop software that's better at it than us.
Guy Reams (06:36.782)
in certain areas, maybe not every area, but in certain areas. Now, is this good or bad? Well, consider the fact that this capability that humans have grown grew completely organically through mitosis, through meiosis, through cell division. We just think about it. Our natural process has led to the formation of an organic functioning brain.
that has developed a consciousness and through that we have inductive reasoning. And that's only one small aspect of the human cognitive ability. There's probably nine or 10 major cognitive abilities the human brain has. So although the machines we develop will undoubtedly surpass us in some areas, it's incredible to think how marvelously imperfect, yet perfect the human mind is. So,
That's actually pretty incredible to think about. So I think I'll chew on this a little bit more today as I eat a turkey sandwich. Thank you.