A.I Morality in Future and Present =NULL;

Today I've asked Copilot (the free version) to calculate something.

The question is: "What, and how?"

The What could result in differing answers even minute by minute (e.g., current exchange rates). If the "how" is not specified somehow, the method used can vary between prompting cycles if alternate methods are available.

GIGO applies, big time, to AI answers. And where variability is not only possible, but likely, one needs to prompt specifically to limit it as much as is possible.
 
The question is: "What, and how?"
The what: A static technical calculation, about converting a cluster to CHS values.
The how: copy/paste of the same prompt.

When I told it about its mistakes, it finally gave me the correct, verifiable answer...
 
Last edited:
So I watched that and here is what I took away from it vs my own reality of watching AI take away a lot of work for me.

AI can be wrong, make bad decisions, have no human touch or watchful eye.

Since mid Jan of 2026, I have watched AI help me with getting referrals. The calls from these calls are always like "I asked ChatGPT and finally it wasn't solving my problem, so I asked it to help me find a person and it listed your name". At least 30 calls now since the beginning of the year.

Then I saw the slow down of calls start in June. My Google analytics started going down. I did research on my own marketing and realized that AI is taking some of my work. Practically any Outlook problem/error is listed in AI and 365 admin "how to" steps. It's getting smarter and smarter, I quit doing some how-to blog posts on my website for just that reason. I'm still marketing with my website, just using more AI prompted keywords so I can get found, rather than solve a problem. I'm adding my personal marketing like my LinkedIn, podcast and youtube to call my website pages so people can see/hear the human.

I have a feeling that people will get sick of reading through 10-20 paragraphs of how to fix something themselves and call again but for now, I'm working on my new 2nd business that cannot be even touched by AI.
 
I have a feeling that people will get sick of reading through 10-20 paragraphs of how to fix something themselves
In recent AI queries I have made on tech issues that have come up, I find one of the most frustrating things is 20 or 30-step answers to any query, with the general format of: #1 If [some action] returns X, then Y, otherwise #2 If [some location] contains Z, then A.....

I have taken to adding "please present any multi-step answers one step at a time" to my queries, but even then, I find they frequently go off on unrelated tangents. Standard searches at least don't do that, but you often have way more time building the path to the desired result youself.
 
I have little doubt that AI used incorrectly will generate at least as much business as it takes away.

All we need to do is look at several recent threads where clients were putting the cart way before the horse, or not even able to give anything approaching a prompt that could generate a valid path for solving anything.

If you can't even semi-precisely define a problem, AI isn't going to get you a viable solution. GIGO writ large.
 
Good 'Prompt Crafting' appears to be a skill that we're going to have to learn well in order to obtain the best information we can from AI agents then!
Just as well I passed my English Language O'Level all those years ago then!
 
In recent AI queries I have made on tech issues that have come up, I find one of the most frustrating things is 20 or 30-step answers to any query, with the general format of: #1 If [some action] returns X, then Y, otherwise #2 If [some location] contains Z, then A.....

I have taken to adding "please present any multi-step answers one step at a time" to my queries, but even then, I find they frequently go off on unrelated tangents. Standard searches at least don't do that, but you often have way more time building the path to the desired result youself.
It's a "Choose your own A.I Adventure" really.
 
All we need to do is look at several recent threads where clients were putting the cart way before the horse,

Agree - I have had a couple of recent queries from folks basically asking for some fully-formed solution to an unstated problem. Dollars to doughnuts these were AI-suggested solutions. My first question whenever I get calls like this is "What problem are you trying to solve?".

Just last week I had a client where we had installed 9 wifi access points to fully cover a large 3-story home & triple garage, plus 2 outdoor areas. Separate VLANs for IOT & Guest traffic & a time-limited kids network. He wanted to kill the IOT network and just put those devices (over 100!) on the regular network. It took several tries, but he finally admitted he had Claude-coded a grafana dashboard to see the status of everything and couldn't get the IOT traffic to report correctly. Incredibly, all I did was teach him how to ask the right question and that "solved" his problem. I put "solved" in quotes because god only knows what security nightmare he has put into place to get this all to work. It was cool, but yikes.

GIGO writ large.

^^This^^ one right here!!
 
Back
Top