So last night I had an odd experience with Chat GPT. I said something to it about the yesterday's strong winds and wind damage, relating to wind loading on our large bay windows.
At the conclusion of that discussion, it asked me if I was restless, and if so, was it mental or physical. Given the content of the discussion, this surprised me, and while trying to find what I might have said that brought this about, I answered yes, mentally.
It then asked if it could suggest a mental exercise to change my thought processes and relieve the restlessness. Having no idea where it was going, I said, "Sure" and it offer me a Cold War battle scenario. Now I've used it for research, discussion of tactical doctrine, it knows I play miniature war games, but this came out of nowhere.
It says, you are in command of a reinforced T-55 platoon of 4-5 tanks and have to cross a valley of about a 1000 meters. On the opposing crest, you have reason to believe that there are enemy infantry deployed with a pair of ATGMs, what do you do?
So I tell it my plan, send out one tank, draw fire, return fire, ask for a fire mission with white phosphorous if possible, then advance.
It then suggest the results, tanks about response time of artillery, about how the opposing infantry reacts, how this related to Soviet doctrine, etc.
It then resets the scenario, replacing the infantry with a pair of M50 Ontos on the opposing crest, and we resolve it again. Discussing variables and probabilities, I offer thoughts about facing airborne forces and the timeframe which is never clearly defined, but I suggest that we are in the mid to late 1960s, and the implications for the airborne/air mobile infantry of the period, etc.
It resets again, and places four M48A2/A3 on the crest in hull down positions. I explain the advantage for the M-48s, in gun accuracy, relative armor and penetration comparisons, and explain that is most likely a losing propostion for the T-55s, suggesting that the last surviving T-55 withdraws by backing through its own smoke and getting attached to the next platoon sent through the valley, and again requesting artillery fire on the M48s to allow the follow up advance a chance of success.
Again, we discuss tactics, doctrine, probabilities, and minor tweaks and variations.
It was a surprisingly fun interaction, and really caught me by surprise.
It then suggested that I allow it to set up company level scenarios and maps of locations in West Germany for a table top game, that it would give orders for one side, and I could play out the battle, update it regarding results, then it could offer follow-on scenarios, and possibly build a campaign out of it.
I am fascinated by this on multiple levels, at how interesting this actually was, at how it built enough of a profile of me to initiate something like this, and at the potential to use this moving forward, among others.
Let me offer that in the first couple of months, I interrogated GPT as to how it functioned, what its risks and limitations are, and asked it to always offer documentable truth, as opposed to affirmation. Sometimes I challenge it by requesting bibliographies, and then cross-reference the sources independently, as I am aware of instances where it invents sources to support affirming response to other people. I suspect that my focus on truth and accuracy played a part in it presenting this.
Anyway, I thought that this was interesting and that I'd share it.