Pages

Thursday, 27 February 2025

Ian Dury’s experiments using ChatGPT

I recently received an email from Ian Dury about his experiments using ChatGPT:

I have started to use ChatGPT to flesh out some of the lower-level details for the campaign and just asked it to produce a scenario for a battle. The results are attached and I thought that they might be of interest. I thought they were pretty good.

This is the instruction I gave to the software:

  • Prepare a fictional scenario for a multi-player wargame set in the Kohat district of the North West Frontier of India in 1881.
  • At a minimum, the scenario should include:
    • Introductory context explaining why the scenario is occurring and the events that triggered it.
    • Details of the forces involved (one or more Pathan tribes, British Indian army troops).
    • An outline map for the scenario.
    • Initial briefings for each player taking command of a force.
    • Victory conditions.
    • Random events (both positive and negative) that could impact any or all of the forces involved.
  • The scenario should be based around an outlying British fort that has been damaged during an earthquake.


He attached a copy of the document that ChatGPT produced … and when I read it, I was impressed. By setting out very specific parameters, Ian ensured that ChatGPT produced results that made a lot of sense and did not include anything that was off piste. The final document is produced below.


Wargame Scenario: The Siege of Fort Lockhart (Kohat, 1881)

Introduction & Context

In the early hours of March 2, 1881, a violent earthquake struck the Kohat region on the North-West Frontier of India. Among the structures affected was Fort Lockhart, a small but vital British outpost guarding the caravan routes and serving as a forward base against hostile tribes. The earthquake severely damaged the outer walls, creating breaches and causing a partial collapse of the south bastion.

Seizing this opportunity, the Orakzai and Afridi tribes have mobilised, gathering warriors for an assault before British reinforcements can arrive. Meanwhile, the British garrison must hold out until relief forces fight their way through the treacherous mountain passes.

The stage is set for a desperate struggle between the determined Pathan tribes and the beleaguered defenders of the British Empire.


Forces Involved

British Garrison at Fort Lockhart

  • Commander: Major Edward Harrington
  • Troops:
    • 1 company of the 5th Punjab Infantry (50 men, professional, well-armed)
    • 1 platoon of the 3rd Sikh Infantry (20 men, experienced and steadfast)
    • 10 Royal Artillerymen operating a single 7-pounder mountain gun
    • 15 Frontier Constabulary troopers (local auxiliaries, variable morale)
    • 10 non-combatants (engineers, medics, and camp followers)
  • Status:
    • Supplies: Limited rations and ammunition (enough for three days of sustained combat)
    • Fortifications: The earthquake has caused breaches in the walls, reducing defensive bonuses.
    • Morale: Generally steady but can be affected by attrition and command decisions.

Tribal Forces

  • Orakzai and Afridi Warriors:
    • Leaders: Malik Zarin Khan (Orakzai) and Gul Rahman (Afridi)
    • Total Strength: Approx. 300 warriors, armed with jezails, swords, and captured British rifles.
    • Tactics: Hit-and-run attacks, ambushes, and massed assaults.
    • Morale: High initially, but dependent on leadership and battle progress.
    • Supply: Largely self-sufficient but will struggle if drawn into a prolonged siege.
  • Possible Reinforcements:
    • More tribal warriors could arrive depending on events.
    • British relief force expected within 3-5 days.

British Relief Force (Optional, Arriving Mid-Scenario)

  • Commander: Colonel Henry Colvin
  • Composition:
    • 1 company of the 21st Punjab Infantry (60 men)
    • 20 cavalry troopers from the 1st Punjab Cavalry
    • 1 mule-drawn 7-pounder mountain gun
    • 10 Pathan scouts and guides
  • Objective: Reach Fort Lockhart before it falls.


Outline Map & Terrain Details

The game master should provide a basic topographical map featuring:

  • Fort Lockhart: A walled fort with damaged sections (indicated on the map), with crumbling bastions and limited visibility from certain towers due to debris.
  • Nearby Hills: Rugged, rocky high ground surrounding the fort, ideal for Pathan snipers and ambushes. The hills provide natural cover and make direct assaults challenging.
  • Approach Roads: Two primary routes exist—one is a narrow, winding mountain path used by the relief force, while the other is a dried-up riverbed frequently used by the tribes for stealthy movement.
  • Stream/Water Source: A small but vital stream lies near the fort, serving as the garrison’s only fresh water supply. The tribes may attempt to cut off access to it.
  • Small Villages and Huts: Scattered along the valley, these provide potential hiding places for Pathan warriors and can be used for staging surprise attacks.
  • Caves and Tunnels: The region is dotted with natural caves, which some tribesmen may use for ambushes or to store extra supplies.
  • Rockslides and Unstable Ground: The earthquake has made parts of the terrain unstable, meaning both sides risk movement hazards, particularly along steep cliffs and loose shale.


Initial Briefings

British Garrison (Major Edward Harrington)

  • Situation: Fort Lockhart has suffered structural damage. The tribes are gathering for an assault. Ammunition and food are limited, and reinforcements are days away. Your duty is to hold the fort at all costs.
  • Orders:
    • Defend the fort: Repair breaches and prepare defensive positions
    • Manage resources: Avoid wasteful expenditure of ammunition.
    • Signal for relief: Use heliographs or runners to coordinate with the approaching column.
    • Handle morale: Keep discipline among your men and prevent desertions.
  • Victory Conditions:
    • Hold the fort for at least four days until reinforcements arrive.
    • Inflict heavy enough losses on the enemy to force a retreat.
    • Successfully evacuate and escape if holding out is impossible.

Tribal Commanders (Malik Zarin Khan & Gul Rahman)

  • Situation: The British fort is weakened. This is your chance to strike a major blow against the imperialists before they receive reinforcements.
  • Orders:
    • Capture or destroy Fort Lockhart.
    • Avoid unnecessary losses. Your warriors are not easily replaced.
    • Cut off British supplies and reinforcements. Use ambushes and harassment tactics.
    • Exploit the terrain. Use sniper fire from the hills and hidden tunnels to weaken the enemy.
  • Victory Conditions:
    • Capture the fort or force its surrender before reinforcements arrive.
    • Destroy the British forces inside, including supplies and weapons.
    • Maintain high tribal morale and avoid heavy casualties.


Random Events Table (Roll 1d6 Each Turn)

  1. Desertion! A group of Frontier Constabulary troopers defect to the tribes.
  2. Sniper Fire. A hidden tribal sniper takes a shot at a British officer.
  3. Delayed Reinforcements. The relief column is delayed by a rockslide.
  4. Earthquake Aftershock! Further damage to the fort; roll for structural collapse.
  5. Surprise Tribal Reinforcements. 50 more warriors arrive to support the attack.
  6. British Airburst Shells. The mountain gun scores a direct hit, breaking an enemy charge.

12 comments:

  1. That is an interesting scenario and more details than I would have expected. Sadly the Random Events are a bit one sided, but that's not difficult to address.

    ReplyDelete
    Replies
    1. WEK 3,

      Whilst it might not be perfect, it shows what is currently possible with the current version of ChatGPT.

      All the best,

      Bob

      Delete
  2. Bob, I have been messing around with both ChatGPT and Grok, seeing what they might do for AI opponents for an operational boardgame. So, far, the results are promising, but it is truly time intensive.

    As AI progresses in ability, we will become spoilt for choices as board and miniature gamers.

    ReplyDelete
    Replies
    1. Justin Penwith,

      I don't think that the situation will remain static with regard to the use of AI in wargaming, and agree that in the not too distant future it will be a very valuable tool for wargamers, especially those of us who do a lot of solo wargaming.

      All the best,

      Bob

      Delete
  3. Bob -
    I'm just wondering if the thing might become a little TOO handy. Here's a thing. I want to play test a modification to the Portable pre-Dreadnought naval rules. Now, I could just shove two fleets together and suck it and see. But I like to build the action into a narrative - or a narrative into the action.

    So, I thought to match my Azurian (French) navy against the Hellenic (Greek). History seems to have provided something of a precedent for such a clash, by the way. At any rate, Azuria and Turcowaz in alliance, the Hellenic kingdom has been showing bellocose indications of support - material support - for Izumrud-Zeleniya, the enemy.

    This could have serious implications for the Aegean Sea lanes from Azuria to Turcowaz. So Azuria embarks upon a 'show of force' to deter Hellenia from entering the lists. A powerful force is sent, ready if necessary to bombard one of the more annoying forts (on Naxos, Chios and/or Lesvos, say), even to take the demonstration to Piraefs itself.

    The Hellenes, not at all averse to a fight, sets out to meet force with force...

    I chose this sort of encounter, with these forces, as being fairly evenly matched. I wonder whether something like ChatGPT would be a help to the imagination or perhaps a substitute... Colour me - for the moment, you understand - sceptical. Mind you, the 'Fort Lockhart' action sounds like a very good and well prepared scenario for even a mini-campaign.

    Cheers,
    Ion

    ReplyDelete
    Replies
    1. Archduke Piccolo (Ion),

      Will using AI become too handy not to use? I think that there is a distinct possibility that it might be. That said, I still think that people with imagination will prefer not to use it and - like you and me - do there own thing.

      I look forward to seeing how your modifications of the rules work … and will no doubt ‘steal’ your best ideas!

      All the best,

      Bob

      Delete
  4. It seems to have produced an interesting scenario, and I can see it being useful for many people who are pressed for time or perhaps needing help to get out of A rut. As for me, I agree with Ion that it could easily become a crutch. Of course, it's easy for me to say as I have never had issues coming up with ideas that are interesting, at least to me.

    ReplyDelete
    Replies
    1. Mark Cordone,

      I think that the scenario is a good one … but Ian Dury’s parameters are such that it gave considerable direction and guidance to ChatGPT, and probably involved more work than a lot of casual AI users would be willing (or able) to put in. I can see myself using something similar if I am stumped one day with regard to developing a scenario and my tired old brain isn’t up to coming up with something. (Funnily enough, that just how I feel today!)

      All the best,

      Bob

      Delete
  5. Reading the comments on this (all of which make sense to me!), I thought it was worth adding some context to why I am using AI tools in this way. My ambition is to run a (probably mostly solo) campaign set in the NW Frontier of India in 1881 - using a version of the map in Featherstone’s ‘Solo Wargaming’ as the basis. For the way I want to run that, I need lots of detail, such as the tribes, their leaders, the relationships between them and their attitude to the British. Similarly, I want lots of potential scenarios that COULD be triggered by events, changes in relationships with or between the tribes, etc. I could create all of that detail myself (there are lots of character generation mechanisms around, or instance) - BUT, it takes a lot of work and, by definition, I will know that detail intimately. Using AI speeds the process massively, adds an element of uncertainty - and means that I may not know all the detail, so can still be taken somewhat by surprise by the ‘programmed’ opponent in a solo war game. Hope that all makes sense!

    ReplyDelete
    Replies
    1. Ian Dury,

      That makes perfect sense … and is - I suspect - one way in which the use of AI by solo wargamers will gradually become more common. It’s certainly one way in which I can envisage myself using it.

      I wonder how long it will be before someone publishes a complete book of AI-generated rules and associated scenarios. I suspect that it won’t be too long!

      All the best,

      Bob

      Delete
    2. Ian and Bob -
      I agree. Ian raises points in connection with uncertainty that I had not considered. I recall playing a certain PC wargame (Steel Panthers) years ago, and a certain scenario requiring the rescue of a stranded German infantry Division from behind enemy lines. Not knowing what was ahead made the first time playing the scenario more fun, more exciting - more engaging -than, say, the tenth.

      So if the AI is capable of springing surprises - plausible ones of course - then that for mine would be a recommendation.

      So is the saving of time and effort when they are at a premium.
      Cheers,
      Ion

      Delete
    3. Archduke Piccolo (Ion),

      Creating a level of uncertainty for a solo player or those taking part in a face-to-face wargame can really increase their enjoyment. I know this from my own experience … and if we can develop ways for AI to enable this to take place in our battles, we can look forward to a new golden age of wargaming. If it also saves us time and effort, then all the better!

      All the best,

      Bob

      Delete

Thank you for leaving a comment. Please note that any comments that are spam or contain phishing messages or that come from Google Accounts that are 'Unknown' will be deleted.