"I just got an AI program to write a set of Napoleonic rules!" Topic

Is this the future? I just got an AI programme to write a set of Napoleonic wargaming rules. And it did a not too bad job – and in just ten minutes!

Even more frightening, it did a better job of writing a posting about these rules on my blog than I could do myself!

Check out the rules and ChatGPT's posting about them here:

link

Pull the plug now! It is another step towards the convergence…

Joking aside, interesting concept. Has anyone fought an AI opponent with miniatures?

….. skynet will achieve self awareness in 10 minutes … 9 …..8 ….

Truly fascinating. Many thanks. Would be interesting if you were to explore this further

I experimented with an AI generator for art recently.

Mixed results. When I asked for squirrels and cockroaches, it just gave me squirrels. It made some cool sci-fi tanks and spacecraft.

The funniest result was when I asked for "woman kissing bearded man." The man and the woman looked great, and they were posed as if thinking about kissing. However, the man's giant beard continued across to the woman's face and ran up to her nose!

That picture of Napoleon is creepy. Is that his mummified face?

I, for one, welcome our new robot overlords.

😅 I can just imagine yellow or red squares around some heads of your figures on the table, a la "Person of Interest..". Samaritan is watching…

Gotta watch those cracks in the table. Lost many a Unit to the void.

Best blog post ever!

ChatGPT even knows the unwritten rules of life.

And here is a list of writing tools, just in case you're staring at a white piece of paper.
One of those AI Buddys may help to overcome the fear.

1.Copy AI : copy.ai
2. Jasper: jasper.ai
3. WriteSonic: writesonic.com
4. ChatGPT3: chat.openai.com
5. Headlime: headlime.com
6. PepperType: peppertype.ai
7. MarkCopy: markcopy.ai
8. Quillbot: quillbot.com
9. Rytr: rytr.me
10. MoonBeam: lnkd.in/ewu28ZVC
11. Simplified : lnkd.in/epc7emeA
12. Lex Page: lex.page
13. Copy Smith: copysmith.ai
14. Subtxt: subtxt.app
15. Ellie Email Assistant: tryellie.com
16. Wordtune: wordtune.com
17. Sudowrite: sudowrite.com
18. Novel: novelai.net
19. Compose: compose.ai
20. Neuroflash: neuroflash.com
21. Yarnit: yarnit.app

It does not follow rules, because it does not know the concept of rules at all, but only generates constructions of probable sentences based on the input it has been given at the prompt.
You cannot play consistent games against it.
It can explain chess to you, but it can't play it, because it doesn't apply rules, but strings together probable symbols, words, sentences, and so on related to chess.

Combine that with a hierarchical memory system, a knowledge graph data structure, an inference logic processor, and let it learn permanently using actual real world data, and you get something really cunning.

Hmph. No, the rules aren't playable, but they're close to playable, and they're short and readable. I've been handed worse once or twice in conventions--and bought one or two from Osprey, come to that.

I will note that none of the "unique mechanisms and special abilities" you mentioned were unfamiliar. I'd actually have called them fairly common. Possibly the AI program has read more Napoleonics rules than you have?

How much do you suppose it wants for an SF skirmish set? Or an adventure story set on pre-Mariner Mars?

I think the AI has read some early Donald Featherstone work.
It promised the stat sheet but did not deliver.
Details were hinted at but not specified.

Actually, Cavcmdr, ChatGPT did give some stats (albeit basic) in the Word file that you can download from the blog.

I wonder if I had drilled down more by asking it specifically for separate stats for British, French, Prussians etc, what it would have come up with?

What amazes me is how the Ai comes up with humour. I thought that was really just a human attribute? I mean, where in the world did it pick up the idea that the size of a unit if you are using plastic soldiers is how many you can fit in your pocket? Did it read this 'joke' somewhere (it probably wasn't Donald Featherstone though, as he used plastic figures properly in his books)? Or did it come up with this all by itself?

@Robert Piepenbrink: I asked ChatGPT to write a short adventure story for you set on pre-Mariner Mars, and this is what it came up with:

Once upon a time, on the planet Mars, a team of brave and intrepid explorers set out on a mission to uncover the secrets of the red planet. Led by the resourceful and clever astronaut, Jack and financed by the President of the United States, who was also known for his love of space exploration and also his love of Space Beer.

As they journeyed through the red sands, they stumbled upon a strange and ancient structure. It was a giant underground chamber, which looked like it had been abandoned for centuries. Being the curious and playful astronaut that he was, Jack couldn't resist exploring the chamber.

However, just as he and his team were about to enter the chamber, the Mariner spacecraft suddenly malfunctioned and fell from the sky, hitting Jack on the head and knocking him unconscious.

When he awoke, he found himself in the chamber. As he and his team explored the chamber, they found themselves surrounded by strange and advanced technology, and also a group of Martians who were all dressed in Hawaiian shirts and sunglasses, sipping on martinis.

The Martians, who were very friendly and welcoming, explained that they were a group of alien party-goers who had been stranded on Mars for eons. They had built this underground chamber as a party lair to have fun and pass the time.

The team was invited to join the Martians for a wild and crazy Martian party that lasted for days. They danced, sang and played games with their new alien friends, and even got a chance to fly in the Martians' spacecraft. Jack and his team also got to sample the famous Martian Space beer that had a kick like a rocket and tasted like intergalactic bliss.

As the party came to an end and the team prepared to leave, the Martians revealed to them that they had been behind the Mariner spacecraft's malfunction, as they didn't want the humans to leave and interrupt their party. But they also gave them the option to stay and join the party forever, but Jack and his team decided to leave and head back to Earth with tales of their wild and hilarious adventure on pre-Mariner Mars.

Upon their return, Jack and his team were greeted as heroes and the President even hosted a dinner in their honor. The President loved the stories of the Martian Space beer and the wild party, he decided to make it the official drink of NASA and also to plan another trip to Mars to join the party with his own team.

And here is a story ChatGPT wrote for me based on the instruction: "write a funny story in English from the perspective of a model French grenadier in a Napoleonic tabletop wargame":

Once upon a time, there was a French grenadier named Pierre. He was a fierce soldier, dedicated to serving his country and his emperor, Napoleon Bonaparte.

As a member of the elite grenadiers, Pierre was a miniature figure in a Napoleonic tabletop wargame. He was carefully painted to match his uniform, and carried a musket, protected by a tall bearskin hat.

One day, Pierre found himself on the battlefield of a Napoleonic tabletop wargame. He and his fellow grenadiers were facing off against the enemy, a group of British redcoats. Each miniature figure was carefully crafted and painted to create an immersive and realistic experience.

The game master called out the commands, and Pierre and his men advanced towards the enemy. The British fired their muskets, but the French grenadiers pressed on, undeterred.

As they closed in on the enemy, Pierre rolled the dice to determine the outcome of the combat. The dice determined the success of the attack and the number of casualties on both sides.

But as the battle raged on, Pierre realized that his musket was just a toy and the bullets were made of plastic. He then decided to use his trusty baguette as a weapon and managed to win the battle by hitting the British general with it.

With the British losing ground, Pierre and his men charged forward, baguettes fixed. They clashed with the enemy in a hilarious melee, but in the end, the French grenadiers emerged victorious.

The game master declared the French the winners, and Pierre and his men cheered in triumph. They had fought bravely and won the day, just as they had done countless times before in the service of their emperor, even if it was just a game.

But as the game ended, Pierre knew that his duty was not yet over. He would continue to fight for his country, in the service of Napoleon, until victory was secured and peace was restored to France, or until he runs out of baguettes.

"I just got an AI program to write a set of Napoleonic rules!"

Now see if you can trick it into painting minis… ;-)

Now see if you can trick it into painting minis

I learned on TMP this would be just a bit more fiddling with the programming of a dumb robot arm in the third dimension.
Lights Out Robot Painting

The smaller KUKA's aren't that expensive.
KUKA Industrial Robot Arms

For certain batch sizes, Chinese toy painting slaves in DongGuang, for example, are cheaper, of course.
Happy Craft

What amazes me is how the Ai comes up with humour. I thought that was really just a human attribute? I mean, where in the world did it pick up the idea that the size of a unit if you are using plastic soldiers is how many you can fit in your pocket? Did it read this 'joke' somewhere (it probably wasn't Donald Featherstone though, as he used plastic figures properly in his books)? Or did it come up with this all by itself?

It doesn't know what humor is, and it doesn't come up with it on its own.
It doesn't "know" anything beside probabilities of text structures.
You enter a sentence at the prompt.
For each character, word, sentence, paragraph, it "knows" a probability distribution of subsequent characters, words, sentences, paragraphs, etc.
It has been trained on billions of texts from the Internet and other sources by calculating these probabilities.
What you interpret as humor is a structure that is common in this hobby in relation to the amount of text.

In a certain sense,ChatGPT is just a mirror.
An unpolished distorting mirror of human interaction via language in text form.
When you're talking to it,you're talking to the babble of humanity.
A lot of times it makes sense, often not.
And yes, you have to put in a lot more work to find sense in it.
Often without result.
It is a text generator, its "intelligence" is the intelligence inherent in human language per se.

But, besides the obvious nonsense, interesting new combinations of text often emerge that can be very inspiring.

For the most part this type of AI technology is more regurgative than generative AI.
But it is a promising approach with a huge potential.

Hmm. Not exactly Leigh Brackett, is it? Still, a step up from Lin Carter, so maybe there's hope. Thanks.

I'd agree with Arjuna overall, but think how many (successful!) authors do nothing original, or gave up years ago and keep recycling the same material. Our AI here has the potential to put old material in new combinations, which is all most of us do.

Where I see real potential is in the underserved niche market. They're writing pastiche Sherlock Holmes faster than I can read it, but if I want a new Patience Campbell McKenna or Philo Vance, things get trickier. Maybe ChatGPT and I can work something out.

Have a look at this.
It's based on an earlier version of the engine ChatGPT is based on.
AIDungeon

AIDungeon – Scenario Search

In the area of visual art, they are a step further already and can choose styles, up to that of individual artists.
On the other hand, it is much more difficult to get good results.
StableDiffusion – On Replicate.com

StableDiffusion – On Huggingface.co

First try on the Replicate version of StableDiffusion with the basic setting and the prompt:
'A wargaming miniature of Napoleon and his Old Guard in Star Wars style'

Crude, and often hilarious, but with some work good results are doable.
My attempts at Bekzinski-style Lovecraftian entities in miniature were pretty impressive, har har.

Try it with ChatGPT, tell him to write like Shakespeare.
It's superficial, but it meets the style of speech.
Maybe you'll have some luck with lesser-known authors.

And always remember that ChatGPT uses everything you wrote in a session and recalculates it together with your new input.
Tell him to switch to other moods, be more precise, give more detailed examples and tell him to imitate them.
At first glance, its results may seem silly and boring.
To use its potential, you need to interact with it.
Also, try some of the other tools I listed above, some already have menus so you can select styles.

It can be great fun but don't expect too much.
In the end, it's your intelligence that counts.
ChatGPT is just an uncalibrated tool for it.

The ChatGPT stories still read better than Harry's autobiography.

The CHatGPT is as good as many human respondents, for example, using a well-worn forum question.

Were the British Guard any good in the Peninsula?

I got
"As a language model AI I do not have personal opinions, but according to history, the British Guard was highly regarded for their discipline, training, and effectiveness in battle during the First and Second World Wars. They were known for their ceremonial duties, but also played a significant role in battles such as the Battle of the Somme and the Battle of Waterloo.

As noted above, the AI has no natural "intelligence" but can quickly string together vaguely related sentences grammatically and correctly.

Tried it with "Explain the 1792 Reglement". It produced a F- junior high school level response where the first and last paragraphs were formulaic blah lazy students copy from Wikipedia. The middle paragraphs were about the French revolution, but all were nonsense wrt the 1792 reglement.

… clearly it has not done the required reading and the essay shows this …

That's freaky!