#187 ChatGPT has Assumptions

The Christian Economist | Dave Arnott

ChatGPT leans liberal and can be deceptive, but Christians should not be afraid of it; It’s simply another machine that will cause creative destruction that makes us economically richer. Toward the end of last semester, I wrote to my Dean in the College of Business, “I either have the best student in the history of our program, or he’s using ChatGPT to write his essay answers.” After noticing some similarities in a few students’ essays, I got something of a sense of what was written by students, and what was written by the Bot. I had asked a strategy question, and the guilty answers contained more financial information than I had asked for. This is not a huge deal. Students have been cheating on answers since Cain said he didn’t know where his brother was. Keeping up with cheating is what we do as teachers and professors. This is simply a new technique. It IS cheating, by the way. Passing off someone else’s work as your own is cheating. Even if the “someone” is a computer program. There are many Biblical warnings about deception. Here’s just one from the Old Testament, and one from the New Testament. Job 15:31 “Let him not deceive himself by trusting what is worthless, for he will get nothing in return”. In 2 Timothy 3:13-14, the apostle Paul warns us, “Evil people and impostors will go from bad to worse as they deceive others and are themselves deceived. But as for you, continue in what you have learned and found to be true because you know from whom you learned it”. So, we shouldn’t be alarmed by a bot trying to change our thinking. Assumptions An article in the Washington Post by Gerrit De Vynck is titled “ChatGPT leans liberal, research shows.” I assume Mr. De Vynck wrote the article, and that he didn’t use AI to write it for him. Now THAT would be interesting, wouldn’t it? Asking a bot to write an article criticizing a bot? He cited a research paper that said the following, “The paper adds to a growing body of research on chatbots showing that despite their designers trying to control potential biases, the bots are infused with assumptions, beliefs, and stereotypes found in the reams of data scraped from the open internet that they are trained on.” So, it’s “garbage in – garbage out.” All computer programs are that way. The frightening thing about this one is its apparent ability to teach itself. That’s reminiscent of the robot Hal’s response in the 1968 movie, 2001 A Space Odyssey, “Sorry Dave, ‘m afraid I can’t do that.” But for now, let’s look back – to assumptions – not forward. The Economics textbook I use in my Macro class at Dallas Baptist University was authored by Gregory Mankiw. He has a section in an early chapter about assumptions. As a warning about false assumptions, I tell this story. I bought this painting while waiting for the Vaporetto – that’s the water bus – in Venice, Italy. In 1609, Galileo hauled the Doge – that’s the governor of the city-state of Venice – up the 323 steps of San Marco tower to show him what he’d found in the night sky, using his newly refined telescope. It seemed to indicate that my predecessors – Catholic Priests – were wrong and that the earth was NOT the center of the solar system. Galileo was tried by my predecessors, who claimed to contain all the information in the world at the time. He tried to defend himself by saying it was only a theory, and he didn’t really believe it. Well, he spent the rest of his life under house arrest. The message was pretty clear: Don’t mess with the folks who held the power of information. Legend has it that after his prosecution, he uttered the famous phrase, “Yet it moves.” You still hear that phrase used occasionally when someone thinks they are clearly correct after losing an argument. Maybe like a reasonable person who does the math and finds that the Inflation Reduction Act actually CAUSED inflation. But I digress. The point is that assumptions matter.