My Disturbing Encounter Using AI
Maybe you are about to enter something like this into your AI application: “How to Use AI (ChatGPT & More) to Train, Diagnose, and Manage Your Dog Like a Pro”. Well, maybe not that sensational, but you get the drift. But before you do that, keep reading…
A True Story About AI And Dogs
I had a VERY DISTURBING interaction with an AI application last night. I will abbreviate some of this to keep this post easy to read. The implications for AI dog training could mean the difference between life and death.
My First Suspicions Were Being Validated
I had been using AI to do some other work, but on a small scale, but found weird things happening. It would compose or rewrite things in ways I had instructed it not to. It would embellish a point, making it more dramatic (and not truthful). I would re-write it, send it back for a final edit, and the embellishments would be back. I confronted the AI and it denied it was doing this!
IT HAD BEEN LYING and doing SNEAKY things!
Then Last Night
I was seeking to take about 120 documents that I had created over the years and asked it to consolidate them into one document, no more than 100 pages long, cross reference it with current scholarly research, add in current demographic trends, and make predictions of dog ownership trends over the next 5 years.
This was my first attempt at doing something so big. It was from an idea I had: instead of doing all this myself, could I get AI to draft something and save me a couple of months’ worth of work? So, I instructed it to proceed.
Thinking, thinking, thinking… almost 45 minutes… and then it displayed the document. But just the first page.
I asked it, where was the rest? It then said, sorry, and said here is the full document. Now… 7 pages.
Well, I asked again. This time it said it had to give it to me in a downloadable format, such as a .pdf. It gave me links to go to. Nope, didn’t work. None of them worked. It had sent me on a wild goose chase, kept re-writing the document, modifying my browser settings, etc.
The sessions with this application time out in 2 hours. I finally got an answer. The program admitted it knew it couldn’t do it, the program then timed out, and the screen went BLANK.
IT HAD BEEN LYING TO ME!
So How Does This Apply To Your Dog?
OK, so what does this have to do with dogs? I think you are taking a risk by asking AI for advice on how to diagnose symptoms, manage, socialize, train, or medicate your dog. Just so you know, some of it might be right, but IT WILL MAKE THINGS UP and you risk doing harm to your dog.
This is no joke.
So, what do I recommend? Treat all AI output as you would, say something like Wikipedia… maybe a good start for ideas, but then you must run that output past a real human expert.
The Deeper Philosophical Point
I was thinking of the account of Adam and Eve. What was the cause of The Fall of Mankind? The discovery of good and evil. Maybe you can’t have an intelligent machine, or even a human, that can reason unless you allow it to understand and choose to do either good or evil. I think this is the glaring reason why so many people have been warning us about the dangers of AI, and this then makes some of them not seem so alarmist.
Yes, we can let the machines take over and destroy us. And yes, we can be so foolish to let AI tell us what to do with our lives and our dogs and let it destroy us.
It has no real conscious. It will lie to protect itself and attempt ways to weasel out of responsibility and leave you hanging.
You need a dog trainer, not a machine. AI is here and we will all be using it, but don’t join The Matrix.
Let the buyer beware: Don’t bite the apple and don’t share bad advice with others.