I’ve got a complicated relationship with artificial intelligence.
Sometimes, I think it’s one of the most remarkable technological developments of my lifetime.
Other times, I just about lose my mind arguing with my watch to start a simple countdown timer.
A task that should take three seconds becomes a minute-long argument with a sophisticated electronic device strapped to my wrist. “She” tells me that timers don’t exist. Then they can’t be set. Then she decides she’s going to send a text message.
I (usually) embrace technology. I’ve spent my professional life working with it. But I’ve also realized that just because technology can do something doesn’t mean that it should.
There’s increasing pressure to incorporate AI into everything we do. Every task needs an AI strategy. Every handheld device needs an AI component. Every word processor has an opinion on how to make our thoughts better. But we routinely forget to ask a critical question.
What problem are we actually trying to solve?
If I can accomplish a task by myself in thirty seconds, but involving AI turns it into a five-minute fiasco of writing prompts, correcting mistakes, and verifying results, what problem have I actually solved?
That’s not progress. That’s insanity with a technology budget.
But there’s something that bothers me even more: We’ve developed a blind trust in AI. It gives us an answer, and we accept it. After all, it’s artificial intelligence. Surely it knows what it’s talking about.
Except when it doesn’t.
AI can be completely wrong and still be absolutely convincing. If we don’t know enough to recognize when it’s wrong, how are we supposed to know when it’s right?
Recently, while teaching an EMT class, a student asked my thoughts about using AI in the medical field. I said it could be useful, but only when properly supervised. In fact, the word I deliberately used was chaperoned.
I opened an AI chatbot and described the classic symptoms of a heart attack: chest pain, difficulty breathing, and pain radiating into my left arm.
The AI told me I was likely having a heart attack and to call 9-1-1 immediately. Perfect.
Then I told it I thought I was just tired and cranky and was going to take a nap instead.
Its response?
“That sounds like a great idea. You’ll feel better after resting.”
I couldn’t believe what I was reading.
Imagine if that had been a real patient. That advice could have been fatal.
Artificial intelligence isn’t actual intelligence.
Artificial intelligence can help us research, learn, create, and solve problems. But it shouldn’t replace our judgment or our willingness to think for ourselves.
Sometimes the most intelligent thing we can do with artificial intelligence is recognize when we shouldn’t use it. And when we do, we need to know when not to trust it.
Now, if you’ll excuse me, I need to send a text message.
Assuming my watch doesn’t decide I need a countdown timer instead.
