God Mode
July 10, 2024 12:12 PM Subscribe
Technically, the returns will diminish because of the nature of LMs: they will return shallow text probabilistically biased toward any religious text in their training corpus. Spiritually, while LMs may marshal text effectively, they can neither “read, mark, learn,” nor “inwardly digest” it. Meditating on divine words is what human beings do in their inner being. This technically cannot and morally should not be automated. Mary could not have outsourced her pondering of the angel’s words to an LM, not only because an LM’s next-item-prediction objective is not pondering, but also because it would have denied those words’ ability to form her. from ChatGPT Goes to Church [Plough]
Note: content is focused on Christianity, Plough is the publishing house of something called the Bruderhof. YMMV
Note: content is focused on Christianity, Plough is the publishing house of something called the Bruderhof. YMMV
Is LM supposed to mean a Language Model, one that might not be a Large Language Model? What's a Small Language Model?
posted by I-Write-Essays at 12:28 PM on July 10 [1 favorite]
posted by I-Write-Essays at 12:28 PM on July 10 [1 favorite]
“ AI chatbot designed to answer questions regarding the Catholic faith, started giving false guidance and even offering one user absolution as a “real” priest, raising questions over the limits of AI in the church setting”
Train that with vast amounts of Monty Python and I’m sure that will go well
posted by cybrcamper at 12:32 PM on July 10 [7 favorites]
Train that with vast amounts of Monty Python and I’m sure that will go well
posted by cybrcamper at 12:32 PM on July 10 [7 favorites]
Calvinistic!
posted by clavdivs at 12:41 PM on July 10 [2 favorites]
posted by clavdivs at 12:41 PM on July 10 [2 favorites]
Calvinballistic!
posted by aubilenon at 12:46 PM on July 10 [15 favorites]
posted by aubilenon at 12:46 PM on July 10 [15 favorites]
I think the article is missing the opportunities for synergy between serving God and using an agile methodology. So I had ChatGPT present an action plan for doing so to leverage the opportunity for empowering a more dynamically responsive religious congregation.
posted by ocschwar at 1:01 PM on July 10 [5 favorites]
posted by ocschwar at 1:01 PM on July 10 [5 favorites]
Train that with vast amounts of Monty Python and I’m sure that will go well
He's not the Messiah, he's a very naughty bot.
posted by ocschwar at 1:02 PM on July 10 [11 favorites]
He's not the Messiah, he's a very naughty bot.
posted by ocschwar at 1:02 PM on July 10 [11 favorites]
At the corner is the store known as Soul Scrolls. It’s a franchise: there are Soul Scrolls in every city centre, in every suburb, or so they say. It must make a lot of profit.
The window of Soul Scrolls is shatterproof. Behind it are printout machines, row on row of them; these machines are known as Holy Rollers, but only among us, it’s a disrespectful nickname. What the machines print is prayers, roll upon roll, prayers going out endlessly. They’re ordered by Compuphone, I’ve overheard the Commander’s Wife doing it. Ordering prayers from Soul Scrolls is supposed to be a sign of piety and faithfulness to the regime, so of course the Commanders’ Wives do it a lot. It helps their husbands’ careers.
There are five different prayers: for health, wealth, a death, a birth, a sin. You pick the one you want, punch in the number, then punch in your own number so your account will be debited, and punch in the number of times you want the prayer repeated.
The machines talk as they print out the prayers; if you like, you can go inside and listen to them, the toneless metallic voices repeating the same thing over and over. Once the prayers have been printed out and said, the paper rolls back through another slot and is recycled into fresh paper again. There are no people inside the building: the machines run by themselves. You can’t hear the voices from outside; only a murmur, a hum, like a devout crowd, on its knees. Each machine has an eye painted in gold on the side, flanked by two small golden wings.
I try to remember what this place sold when it was a store, before it was turned into Soul Scrolls. I think it was lingerie. Pink and silver boxes, coloured pantyhose, brassieres with lace, silk scarves? Something lost.
Ofglen and I stand outside Soul Scrolls, looking through the shatterproof windows, watching the prayers well out from the machines and disappear again through the slot, back to the realm of the unsaid. Now I shift my gaze. What I see is not the machines, but Ofglen, reflected in the glass of the window. She’s looking straight at me.
We can see into each other’s eyes. This is the first time I’ve ever seen Ofglen’s eyes, directly, steadily, not aslant. Her face is oval pink, plump but not fat, her eyes roundish.
She holds my stare in the glass, level, unwavering. Now it’s hard to look away. There’s a shock in this seeing; it’s like seeing somebody naked, for the first time. There is risk, suddenly, in the air between us, where there was none before. Even this meeting of eyes holds danger. Though there’s nobody near.
At last Ofglen speaks. “Do you think God listens,” she says, “to these machines?” She is whispering: our habit at the Centre.
In the past this would have been a trivial enough remark, a kind of scholarly speculation. Right now it’s treason.
I could scream. I could run away. I could turn from her silently, to show her I won’t tolerate this kind of talk in my presence. Subversion, sedition, blasphemy, heresy, all rolled into one.
I steel myself. “No,” I say.
She lets out her breath, in a long sigh of relief. We have crossed the invisible line together. “Neither do I,” she says.
posted by AlSweigart at 2:20 PM on July 10 [22 favorites]
The window of Soul Scrolls is shatterproof. Behind it are printout machines, row on row of them; these machines are known as Holy Rollers, but only among us, it’s a disrespectful nickname. What the machines print is prayers, roll upon roll, prayers going out endlessly. They’re ordered by Compuphone, I’ve overheard the Commander’s Wife doing it. Ordering prayers from Soul Scrolls is supposed to be a sign of piety and faithfulness to the regime, so of course the Commanders’ Wives do it a lot. It helps their husbands’ careers.
There are five different prayers: for health, wealth, a death, a birth, a sin. You pick the one you want, punch in the number, then punch in your own number so your account will be debited, and punch in the number of times you want the prayer repeated.
The machines talk as they print out the prayers; if you like, you can go inside and listen to them, the toneless metallic voices repeating the same thing over and over. Once the prayers have been printed out and said, the paper rolls back through another slot and is recycled into fresh paper again. There are no people inside the building: the machines run by themselves. You can’t hear the voices from outside; only a murmur, a hum, like a devout crowd, on its knees. Each machine has an eye painted in gold on the side, flanked by two small golden wings.
I try to remember what this place sold when it was a store, before it was turned into Soul Scrolls. I think it was lingerie. Pink and silver boxes, coloured pantyhose, brassieres with lace, silk scarves? Something lost.
Ofglen and I stand outside Soul Scrolls, looking through the shatterproof windows, watching the prayers well out from the machines and disappear again through the slot, back to the realm of the unsaid. Now I shift my gaze. What I see is not the machines, but Ofglen, reflected in the glass of the window. She’s looking straight at me.
We can see into each other’s eyes. This is the first time I’ve ever seen Ofglen’s eyes, directly, steadily, not aslant. Her face is oval pink, plump but not fat, her eyes roundish.
She holds my stare in the glass, level, unwavering. Now it’s hard to look away. There’s a shock in this seeing; it’s like seeing somebody naked, for the first time. There is risk, suddenly, in the air between us, where there was none before. Even this meeting of eyes holds danger. Though there’s nobody near.
At last Ofglen speaks. “Do you think God listens,” she says, “to these machines?” She is whispering: our habit at the Centre.
In the past this would have been a trivial enough remark, a kind of scholarly speculation. Right now it’s treason.
I could scream. I could run away. I could turn from her silently, to show her I won’t tolerate this kind of talk in my presence. Subversion, sedition, blasphemy, heresy, all rolled into one.
I steel myself. “No,” I say.
She lets out her breath, in a long sigh of relief. We have crossed the invisible line together. “Neither do I,” she says.
posted by AlSweigart at 2:20 PM on July 10 [22 favorites]
Two years ago previously on mefi: I like to talk (Google engineer Blake Lemoine fired for claiming Google's AI was alive.)
posted by AlSweigart at 2:24 PM on July 10 [1 favorite]
posted by AlSweigart at 2:24 PM on July 10 [1 favorite]
Is LM supposed to mean a Language Model, one that might not be a Large Language Model? What's a Small Language Model?
I can't tell if this is a joke but Natural Language Processing (NLP) [WP] has been a part of AI pretty much from the beginning. Prior to to 2017's Attention Is All You Need [WP], softmax and the transformer architecture, there were a large number of techniques for NLP some of which involved far smaller neural networks like Word2Vec and some of which were just hard-coded heuristics like TFIDF. What distinguishes modern language models as "large" - versus earlier, smaller neural network approaches - is that training is largely unsupervised, and they rely on transfer learning to apply rules or patterns to completely new prompts or text (read: zero shot).
posted by Ryvar at 2:34 PM on July 10 [4 favorites]
I can't tell if this is a joke but Natural Language Processing (NLP) [WP] has been a part of AI pretty much from the beginning. Prior to to 2017's Attention Is All You Need [WP], softmax and the transformer architecture, there were a large number of techniques for NLP some of which involved far smaller neural networks like Word2Vec and some of which were just hard-coded heuristics like TFIDF. What distinguishes modern language models as "large" - versus earlier, smaller neural network approaches - is that training is largely unsupervised, and they rely on transfer learning to apply rules or patterns to completely new prompts or text (read: zero shot).
posted by Ryvar at 2:34 PM on July 10 [4 favorites]
"Small" language model is also sometimes used to refer to models with only a few billion parameters but the same transformer architecture.
posted by Pyry at 3:18 PM on July 10 [2 favorites]
posted by Pyry at 3:18 PM on July 10 [2 favorites]
Why shouldn't the machines or their owners get in on the grift?
posted by drewbage1847 at 3:43 PM on July 10 [1 favorite]
posted by drewbage1847 at 3:43 PM on July 10 [1 favorite]
So, I have been to a Buddhist temple where there was a small building where one could enter, take ahold of the spoke of a cabinet full of sutras mounted in a capstan, and turn it through a revolution before exiting. We were assure this would give us all the merit of having read those sutras. Other temples had cylinders marked with prayers which one could walk by and turn to generate prayers and merit. I have my doubts, but it’s a tradition.
posted by GenjiandProust at 3:48 PM on July 10 [7 favorites]
posted by GenjiandProust at 3:48 PM on July 10 [7 favorites]
something something server farm something heavenly host
posted by cortex at 4:07 PM on July 10 [2 favorites]
posted by cortex at 4:07 PM on July 10 [2 favorites]
Overhead, without any fuss, the stars were going out.
posted by SPrintF at 4:32 PM on July 10 [7 favorites]
posted by SPrintF at 4:32 PM on July 10 [7 favorites]
Don't let me stay, don't let me stay
My logic says burn, so send me away
Your minds are too green, I despise all I've seen
You can't stake your lives on a Saviour Machine
posted by JohnFromGR at 4:34 PM on July 10 [1 favorite]
My logic says burn, so send me away
Your minds are too green, I despise all I've seen
You can't stake your lives on a Saviour Machine
posted by JohnFromGR at 4:34 PM on July 10 [1 favorite]
Amen
posted by blue shadows at 1:25 AM on July 11 [1 favorite]
posted by blue shadows at 1:25 AM on July 11 [1 favorite]
I think the article is missing the opportunities for synergy between serving God and using an agile methodology. So I had ChatGPT present an action plan for doing so to leverage the opportunity for empowering a more dynamically responsive religious congregation.
I was hoping for something insane and amusing, but this is just straight out of an exceptionally boring Evangelical self-help book.
posted by straight at 7:02 AM on July 11 [2 favorites]
I was hoping for something insane and amusing, but this is just straight out of an exceptionally boring Evangelical self-help book.
posted by straight at 7:02 AM on July 11 [2 favorites]
Or an employee manual for an Evangelical church. They increasingly replace the idea of a denomination with that of a Brand (TM). One near my house is called Edify. And the brands are franchised like fast food.
posted by ocschwar at 2:21 PM on July 11 [1 favorite]
posted by ocschwar at 2:21 PM on July 11 [1 favorite]
« Older Government flags massive expansion of Antarctic... | I hope I’m never loud and lonely enough to want to... Newer »
[wiki:] tayus ex machina
posted by HearHere at 12:27 PM on July 10 [4 favorites]