Kevin Wilson
@kevinwilson42.bsky.social
Boardgame designer best known for Arkham Horror, Descent: Journeys in the Dark, A Gentle Rain, and the Kinfire series of games. Now Director of Game Design at Incredible Dream Studios. He/him.
I really should go back and finish that one. It was a good time, but I got distracted like I so often do these days with longer games and didn't get back to it.
November 26, 2025 at 4:45 AM
I really should go back and finish that one. It was a good time, but I got distracted like I so often do these days with longer games and didn't get back to it.
Been a little while since I baked a pound cake, but I'm confident I can pick that recipe back up at any time. After baking 20 or 30 of them I feel pretty proficient at it.
November 22, 2025 at 7:49 AM
Been a little while since I baked a pound cake, but I'm confident I can pick that recipe back up at any time. After baking 20 or 30 of them I feel pretty proficient at it.
I really like it when I start watching a "simple recipe for 1 or 2 people" and the person immediately lays out an array of like 20 ingredients. That's when I usually laugh and tell youtube to banish them to the shadow realm.
November 22, 2025 at 5:57 AM
I really like it when I start watching a "simple recipe for 1 or 2 people" and the person immediately lays out an array of like 20 ingredients. That's when I usually laugh and tell youtube to banish them to the shadow realm.
Even as the world struggles to meet emissions goals, this horsecrap is churning out pollution to create nonsense slop in order to save rich people a nickel. At its core, it's a Mechanical Turk, just another scam like crypto, blockchain, and NFTs. AI has real, genuine uses, but this is none of that.
November 21, 2025 at 12:06 AM
Even as the world struggles to meet emissions goals, this horsecrap is churning out pollution to create nonsense slop in order to save rich people a nickel. At its core, it's a Mechanical Turk, just another scam like crypto, blockchain, and NFTs. AI has real, genuine uses, but this is none of that.
LLMs as they are represent a malignant use of the tech, and are a theft engine for the wealthy. Yet another attempt to sidestep copyright law and avoid paying people for their creations. They're being used to drive wages down in a number of industries and are obscene wastes of energy.
November 21, 2025 at 12:06 AM
LLMs as they are represent a malignant use of the tech, and are a theft engine for the wealthy. Yet another attempt to sidestep copyright law and avoid paying people for their creations. They're being used to drive wages down in a number of industries and are obscene wastes of energy.
That's why it reads like 10,000 dead-eyed corpos even when it gives you a recipe for poison instead of chicken soup. So what I'm saying here is, be really careful before you start trusting the crap spewing out of the nonsense engine. And don't fool yourself into thinking it creates anything.
November 21, 2025 at 12:06 AM
That's why it reads like 10,000 dead-eyed corpos even when it gives you a recipe for poison instead of chicken soup. So what I'm saying here is, be really careful before you start trusting the crap spewing out of the nonsense engine. And don't fool yourself into thinking it creates anything.
The reason everything that LLMs spew out is a bizarre mix of super blandness and incoherent nonsense is because of the things I've explained above. It's an averaging machine with no awareness or real intelligence, and endless human hours have been poured into trying to create basic fencing for it.
November 21, 2025 at 12:06 AM
The reason everything that LLMs spew out is a bizarre mix of super blandness and incoherent nonsense is because of the things I've explained above. It's an averaging machine with no awareness or real intelligence, and endless human hours have been poured into trying to create basic fencing for it.
If you give a chatbot enough hard-coded responses, you can make it seem fairly clever, but anything outside that expected set of inputs still coughs up an error. It can't really improvise a response outside of what its code contains. It can't think, it can't create, and it's not your friend.
November 21, 2025 at 12:06 AM
If you give a chatbot enough hard-coded responses, you can make it seem fairly clever, but anything outside that expected set of inputs still coughs up an error. It can't really improvise a response outside of what its code contains. It can't think, it can't create, and it's not your friend.
Natural language models have been around for awhile as well in various forms, such as ELIZA in the 60's which could perform a very vaguely passable imitation of a therapist by reflecting your statements back at you. "Why do you think that <thing you just typed>?"
November 21, 2025 at 12:06 AM
Natural language models have been around for awhile as well in various forms, such as ELIZA in the 60's which could perform a very vaguely passable imitation of a therapist by reflecting your statements back at you. "Why do you think that <thing you just typed>?"
If you went out onto the internet and started posting many thousands of pictures of spiders and labeling them "cat", this would start to poison the LLM concept of "cat" and you'd start getting hideous catspiders every time someone asked for a picture of a "cat". The LLM wouldn't know the difference.
November 21, 2025 at 12:06 AM
If you went out onto the internet and started posting many thousands of pictures of spiders and labeling them "cat", this would start to poison the LLM concept of "cat" and you'd start getting hideous catspiders every time someone asked for a picture of a "cat". The LLM wouldn't know the difference.
So, they've taken that basic tech and combined it with a few other things to create an averaging machine. It looks for many pictures of "cat" and averages them together in various ways to create a "new" image of a cat, but it has no real understanding of the concept of a cat beyond the word.
November 21, 2025 at 12:06 AM
So, they've taken that basic tech and combined it with a few other things to create an averaging machine. It looks for many pictures of "cat" and averages them together in various ways to create a "new" image of a cat, but it has no real understanding of the concept of a cat beyond the word.
They almost certainly use a variation of a tech called neural networking, which was used in computers like Deep Blue for chess playing. Basically, links between different states (or letters/words/whatever) are weighted with desirability. So, "The" being followed by "End" might be highly rated.
November 21, 2025 at 12:06 AM
They almost certainly use a variation of a tech called neural networking, which was used in computers like Deep Blue for chess playing. Basically, links between different states (or letters/words/whatever) are weighted with desirability. So, "The" being followed by "End" might be highly rated.
I learned two designations of AI - hard and soft AI. Hard AI was a true thinking, self-aware agent that could have original ideas. Soft AI was simply the ability to imitate intelligence, which given the limited understanding of our own intelligence, is a much more modest goal. LLMs aren't even that.
November 21, 2025 at 12:06 AM
I learned two designations of AI - hard and soft AI. Hard AI was a true thinking, self-aware agent that could have original ideas. Soft AI was simply the ability to imitate intelligence, which given the limited understanding of our own intelligence, is a much more modest goal. LLMs aren't even that.
It doesn't and cannot think. It's not even as smart as a parrot, which can at least learn that when it says "cracker" it'll get a nice little treat. That's to say nothing of the idea of second order thought - the idea that you can think about the act of thinking. That's not even in the ballpark.
November 21, 2025 at 12:06 AM
It doesn't and cannot think. It's not even as smart as a parrot, which can at least learn that when it says "cracker" it'll get a nice little treat. That's to say nothing of the idea of second order thought - the idea that you can think about the act of thinking. That's not even in the ballpark.
And honestly, for LLMs, not much looks to have changed, aside from the marketing, computing speed, and access to very large data sets. But the thing is, it's all basically sparkling autocorrect at its heart. It looks through its data set to see what words appear most often in what order.
November 21, 2025 at 12:06 AM
And honestly, for LLMs, not much looks to have changed, aside from the marketing, computing speed, and access to very large data sets. But the thing is, it's all basically sparkling autocorrect at its heart. It looks through its data set to see what words appear most often in what order.
I took programming classes as well as psychology, philosophy, neurobiology, linguistics, and a few other things in addition to some folklore and mythology that I took because I was interested in it. So, I had a pretty decent idea of what the state of the art of AI was then, and what it was used for.
November 21, 2025 at 12:06 AM
I took programming classes as well as psychology, philosophy, neurobiology, linguistics, and a few other things in addition to some folklore and mythology that I took because I was interested in it. So, I had a pretty decent idea of what the state of the art of AI was then, and what it was used for.
Oh sure, I doubt it too, but I can't say it feels like an unreasonable response to me.
November 20, 2025 at 11:06 PM
Oh sure, I doubt it too, but I can't say it feels like an unreasonable response to me.
Many creatives' takeaway from the last decade has been, "Don't be even a little subtle. A lot of people can't handle it." I mean, I thought it was pretty obvious the Empire was the baddies in Star Wars, and yet...<shrugs>
November 20, 2025 at 10:41 PM
Many creatives' takeaway from the last decade has been, "Don't be even a little subtle. A lot of people can't handle it." I mean, I thought it was pretty obvious the Empire was the baddies in Star Wars, and yet...<shrugs>
Peoples' giveashit dried up and blew away in the face of their mild inconvenience. The mines were always gonna reopen too soon, but the pushback on masks was literally a mask-off moment for many, many people.
November 20, 2025 at 10:38 PM
Peoples' giveashit dried up and blew away in the face of their mild inconvenience. The mines were always gonna reopen too soon, but the pushback on masks was literally a mask-off moment for many, many people.
What a weird experience.
November 20, 2025 at 9:07 PM
What a weird experience.