There’s been an endless amount of talk about AI in recent years. People are worried about what jobs it might take over, and we’ve seen instances of people taking their conversations with AI too far, including psychosis, drug abuse, and suicide.
We’ve also seen plenty of discussion about the benefits of AI. A couple of my PJ Media friends and colleagues have ingenious uses for it. Sarah Anderson uses ChatGPT to help her organize her home projects, while Stephen Green has been using it to help him devise plot points and characterizations for a novel he’s considering working on — something I’d like to try, too.
“When I first started this experiment, I basically told it I was overwhelmed and had too much on my plate and felt like I was turning in circles trying to rectify it,” Sarah told me. “We went through my life over the last five or so years and it came up with what’s wrong and what’s right and what I can do about it, and it’s working splendidly.”
I use Grok for research and ChatGPT for punchy headlines. I’ve even used ChatGPT’s image generators to help me envision my weight loss goals and test drive different beard lengths and haircuts.
When Steve, Sarah, and I (and countless others) use AI tools, we’re not under the illusion that we’re talking to human beings. I have hard data to back this up, but I think that generally the people who fall under the spell of AI to their detriment do buy into the notion that they’re interacting with someone real.
A little diversion: I’ve mentioned before that I’ve been watching “Knots Landing,” and one of the subplots in the show’s 10th season (1988-89) involves two college students who are developing security software. It was strange to hear them talk about “artificial intelligence” while also talking about backing up data on floppy discs.
We also shouldn’t believe that AI has all the answers or information; instead, it’s regurgitating data it finds elsewhere. This is why we saw liberal bias in Grok when it came to the Trump administration’s National Weather Service budget cuts. Grok was spitting back out the same left-wing misinformation that it found in mainstream media articles about the budget.
Recommended: Rockies vs. White Sox: 'The World Series for the Two Worst Teams'
AI isn’t reasoning, either. Shane Morris explained on Monday’s “Breakpoint” podcast:
A groundbreaking new study from Apple entitled “The Illusion of Thinking” showed this by subjecting AI models to various challenges designed to test for reasoning ability. Using logic puzzles of increasing complexity, the researchers found that even today’s most advanced AIs didn’t understand or solve problems, but merely pattern-matched.
Rather than learn from or extrapolate solutions as a genuinely intelligent entity might, AI “reasoning models” gave up when problems got too complex, experiencing “complete collapse,” no matter how much computing power researchers gave them.
This was true even when the AIs were given explicit algorithms to follow.
No computer is powerful enough to match the thinking power of the human brain — at least not anytime soon. This means that AI is only as good as the people using it. I can see that in the prompts that I use. If I don’t phrase things just right, I’m liable to get useless or inaccurate results.
This principle applies on moral grounds, too. If we’re not grounded in morality, even when we’re using AI, we’re not going to get results that reflect virtue. Of course, I see things from a Christian worldview, so that’s the framework from which I approach my interactions with people and computers. People who have no ethical framework whatsoever won’t approach AI with a moral grounding, which can lead them down some terrible rabbit holes.
As AI becomes ever more ubiquitous, it serves us well to remember that it’s never going to be as good as its users. That’s as important when we hear the horror stories as it is when we discover how beneficial AI can be.
