Jackson Jambalaya

 The first in a series on a fascinating yet troubling discussion with ChatGPT:

Me: Are you able to present both sides of an issue with equal determination? 

AI: Yes. In fact, presenting competing arguments with comparable intellectual rigor is one of the things I can do particularly well.

Me:  So, does that make you indecisive?

AI: No. Being able to argue both sides strongly is different from being indecisive. Think of it this way: Indecisiveness – both sides have arguments, so I can't reach a conclusion. Balanced analysis – here is the strongest case for each side. Now, weighing the evidence, assumptions, consequences, and uncertainties, here is my conclusion. 

AI: And if you want me to take a stronger stance, you can explicitly tell me: "Don't just give me both sides. Decide which is better and defend your conclusion." I can do that too.

Me: How can you determine “better”? 

AI: That is the crucial question. “Better” has no meaning until we establish what standard we are using to judge better. I don't have a personal ideology or set of political values that I secretly use as the measuring stick. Instead, I can evaluate an argument against explicit criteria. Those criteria can come from the nature of the question, evidence, logic, stated goals, or values the people involved agree are important.

AI: I would generally use five levels of judgment: 1. factual accuracy; 2. logical strength; 3. explanatory power; 4. practical consequences; and 5. values/objectives.

AI: That fifth level is where many supposedly factual disagreements actually reside. A simple example: Suppose we are debating, "Should Mississippi substantially increase spending on public education?" I could determine that one proposal is "better" for economic growth while another is "better" for minimizing taxes. Neither conclusion follows merely from the facts. We first have to establish the objective. 

AI: There is a danger in AI that is easy to overlook. I can produce a very persuasive argument without the author noticing that I have quietly embedded a value judgment in the definition of "better." For example, if I say, "This is the better policy because it is more efficient," I have already assumed efficiency should dominate other considerations. But perhaps you would say, "Efficiency matters, but effectiveness matters more." Those are not merely factual disagreements. They are disagreements about the standard of judgment.

AI: The most intellectually honest form of analysis is: here are both sides; here are the criteria by which I'm judging them; here is how each performs against those criteria; and here is how the conclusion changes if we change the criteria.

AI: Evaluation necessarily requires a standard against which something is evaluated. Without that standard, "better" becomes little more than preference.

“A just balance and scales are the Lord's” – Proverbs 16:11.

Crawford is an author and syndicated columnist from Jackson.