Rethinking My Views on AI

Let me be honest: I’ve never been a big fan of artificial intelligence (AI). My reasons for this are varied.

In part, I just don’t trust it to be accurate. All of the models, after all, are trained on all of the content on the web and there is no doubt that there’s a ton of inaccurate and downright wrong (deliberately and accidental) information out there. Given that it is trained on at least a fair amount of wrong information it would be foolish to trust that the responses produced by AI would be any more correct. It’s the old “Garbage in, garbage out” idea.

Another reason I’m not a fan is that I don’t trust that the providers will do anything to protect my privacy. People who share their personal, private information, e.g. financial records or health information, are essentially giving that information to the world.

Not a good plan.

Perhaps my general feeling for AI would be different if it had not just suddenly popped onto the scene and wasn’t being pushed so heavily by everyone. Maybe I’m just paranoid or conspiracy minded, but there just seems to be something suspicious to me that all of these software companies are pushing AI into their products so rapidly. That’s a lot of work and expense to go to unless there’s a financial payoff for them somewhere. I suspect that “somewhere” is selling my data. And I’m just not on-board with that.

As an educator and an autodidact, another reason I’m not fond of AI is that too many people use it to short circuit their need to think, to learn, or to remember. I like learning. I like thinking. And I think that more people need to do more of both. There’s already a dearth of people who actually think through a situation or scenario critically. All you have to do is pay attention to realize how true that is.

It’s important, in my mind, that people actually learn rather than rely on outside sources for everything they do. I fear that at some point we’ll end up unable to deal with issues without the help of computers because we’ll lack the foundational knowledge needed to actually think things through. And what happens when those computers are not available?

When I was teaching medical math, I had students who argued that their calculators could be programmed to do the math that was needed. IV pumps can do the calculations for drug dosing, and so on. And that’s true. But what about when the power is out for extended periods of time? The calculators’ batteries will eventually fail, the IV pumps won’t work. What then? You need to know how to do the calculations so that you can safely care for your patient. And, what if you put in the wrong values? Will you recognize an improbable result? (The answer to that question was always “no”. Student’s will give me answers that simply defied reality, delivering four or five times the amount available.)

But there are benefits

Now that I’ve dissed AI, let me backpedal just a little bit.

Recently I was working on a networking project that required knowledge of firewalls that I didn’t have. I spent some time trying to learn and figure out what was needed but the complexity of what I was trying to accomplish was beyond my skillset. After struggling with the problem for a day or two, I finally decided to try AI. I plugged in all the information I could, just as I would describe it to a friend, and asked Claude to provide the necessary settings. But I didn’t just want to steps, I wanted explanations of everything so that I could actually learn from the process. That is, don’t just tell me what to do tell me why I’m doing it.

And it did.

And the steps provided worked perfectly. Within an hour I had a functioning network with all of the requirements I needed perfectly satisfied. There were a few glitches but those were because I didn’t specify a couple of parameters. Because I had a basic understanding of what I was trying to do, I was quickly able to figure out those corrections. I didn’t need to rely on Claude to do that. I was learning.

Unfortunately, my sense is that most people don’t use AI to learn or to clarify a complex problem; they use it as a shortcut to a goal. And that goal is not to learn. That includes students who use AI as a shortcut to a grade. In school grading is intended to provide feedback on how well you understand a concept. But that’s not how most people (adults and students) view it. That’s a different discussion for another day.

All of that said, I still don’t think that I’ll use AI on any regular basis. I still prefer to do things the old-fashioned way – by learning, thinking, and problem solving. But, I do recognize that there are situations where the challenges are sufficiently complex or exceed my capabilities such that using AI is useful to solve the immediate problem while teaching me the “why” behind the solution. So long as I’m learning, then I don’t think I’ve lost anything.