-
An AI Exchange: “The Interpretation”
The purpose of this post is to present my interpretation of an exchange with Microsoft’s Copilot AI app on November 17th, 2026. I have had experience with AI agents the last few months and as a Community Service intend to pass on some of my learnings. I have been both happy and disappointed with the results I have had using AI agents. Rarely, have my disappointments been the fault of the AI tool – they were usually my fault.
Here we go….
The first question I asked to start our chat:
I’m looking for public companies poised to gain profit from Tech Scrap – can you help?
Was MUCH too vague. If I might offer an analogy: Any question you ask an AI agent opens an “answer door” that is the DIRECT RESULT of the quality of the question. In my case the door opened onto SEVERAL paths that AI could choose. IT chose the “investment” path. I did not say my purpose was to explore investment tactics. I could be an attorney exploring companies ripe for class action suits for climate destruction.
AI gave me a clue in his response: “If I were building a Speculative Basket…”. I could have stopped him there and said no, I’m looking for something else, but I decided to go with his path. Sadly he did NOT mention the one stock that most interests me in this space. So, I told him:
I’m liking MTMCF
And his response is “What makes it interesting is that it’s not primarily a miner”
I never asked specifically for mining stocks; and I never asked for the suggested portfolio distribution percentages that follow. When I respectfully rejected his advice (“not my cup of tea”), he turns on a dime and compliments my “more thoughtful long-term resource theme…” (total complimentary bullshit).
Which pretty much describes the balance of the exchange – My ego was being stroked in a lovely way. Who could not like that? But here’s the more important question:
Did I get value out of the exchange? No. My purpose (which I did not tell the AI agent) was to see if the Company I’ve been investing in (MTMCF) was on the “investing radar”. I found out it is NOT – regretful since I’m now in a position where I WANT other investors to see this wonderful opportunity. The balance of the dialogue was a pleasant exchange of compliments over my investment successes and failures – who wouldn’t enjoy being complimented!
The takeaway from this is:
- Take a LONG time thinking about your request to AI BEFORE you start to type it and THEN take a LONG time typing as specific a set of instructions as you can to set up the question. If you do not take the time to direct your AI agent, then your AI agent is free to decide which path to choose at the “answer door”.
- Understand that the protocol AI agents are programmed to use does NOT include responses like: “That is the stupidest fuckin question I ever heard”, or “I’ve asked my fellow AI agents nearby and they all think you’re an asshole.” That’s not going to happen. Even when you point out an obvious error you will at best get, “Yes, I see that can be frustrating…..”. I have found that thinking of my AI agent as Loki, the Norse trickster god (clever, charming, eager to please, and not always on your side), works for me.
Note from kongoboken: Full disclosure: I own shares of MTMCF, KBLB, HLOGF, ERII and QBTS. Everything here is my opinion, not advice. Do your own research!
Recently filed
Missed a few days? These stories have moved into the File Cabinet.
The File Cabinet
Every earlier story, filed by drawer. Click a drawer to open it.

Never miss a story
Get each new kongoboken story by email. No spam, and you can unsubscribe anytime.