Yes, this is the first .va link submitted. And before this thread there was only one link to a .va domain in a comment, to press.vatican.va. The story was about invalid Catholic baptisms and I removed it for being off-topic.
He has a scientific background. In 1977 he graduated from Villanova University (Philadelphia). His field? Mathematics. This is more sci-fi than the document itself.
Comparatively, previous Popes have rarely ever made such sweeping declarations w.r.t. (then) modern events.
What?
Popes have consistently issued encyclicals and other formal declarations addressing contemporary issues throughout history. This isn't even the first official Vatican document addressing artificial intelligence, the doctrinal note Antiqua et nova was issued by the Holy Office at the beginning of last year. This very document's opening chapter walks through the lineage of papal engagement with contemporary issues. And there are a great number of papal encyclicals dealing with issues of their times. Going off the top of my head here: Sublimis Deus on the status of indigenous Americans; Rerum novarum on the conditions of workers, capital, and labour; Quadregesimo anno on, among other things, dangers of unrestrained capitalism, socialism, and communism; Mit Brennender Sorge against nazism, written specifically in German to be smuggled into the Reich and read off the pulpits; Pacem in Terris on nuclear war in the wake of Cuban missile crisis; Humanae Vitae on contraception, directly responding to the invention of the birth control pill and sexual revolution; Laudato Si on climate change; Fratelli Tutti framed around the covid pandemic.
And also, just as a note, an encyclical is an authoritative teaching document, nothing less and nothing more. It's not a switch that aligns 1.4 billion of us Catholics behind a single objective.
Oof, thank you for the correction. The news source I originally read this from framed this very differently, which gave me a rather uninformed opinion. That source shall be removed from rotation, and I've deleted the comment to avoid misinforming others, too.
an encyclical is an authoritative teaching document
I'd be curious to hear more about this. Do you have a good resource that you'd recommend on how these are actually implemented downstream? Is it meant to be "you should teach this ASAP" or "if asked, this is the authoritative answer"?
Broadly speaking, AIUI it's the position that the Pope (Catholic Bishop of Rome) has decided to take on something, taking into account existing church teaching and the expressed position of his predecessors (hence much of the verbosity), and also the personal experience and education of the writer.
It's intended to inform and teach other bishops of the Roman Catholic Church, and will normally also be circulated to bishops of the other Catholic Churches who will respect the Bishop of Rome even if they do not acknowledge his supremacy.
It does not of itself necessarily establish doctrine and law, but might be a prelude to something that does.
I'm not sure if "implemented downstream" is the correct framing, because an encyclical is not a law or procedural guideline. It's perhaps a little closer to what a scientific paper might be. One might not necessarily see a scientific paper directly implemented anywhere, but practical applications might be derived from it by means of engineering. It's something of a similar situation here: it's unlikely that the encyclical will immediately and directly command some kind of action with a specific deadline, but rather will over time diffuse into more everyday material such as homilies, pastorals, RCIA instruction, theology courses, &c. that refer to it.
"Teaching authority" in this case refers to a somewhat nebulous concept of Catholic magisterium (rather than teaching in the sense of education), which broadly means the authority of the Church to interpret scripture and tradition. This idea might be quite unfamiliar to those growing up in any kind of protestant-adjacent environment, which treats scripture as the final authority, with the believer expected to engage and interpret it directly. The encyclical genre itself is unusual in that it's pitched simultaneously to Catholics (where it carries magisterial weight) and to "all people of good will" (where it functions as a moral argument that has to stand on its own merits). Laudato Si is a good example of this: alongside inspiring various internal initiatives to promote ecology in the Church it also served as a reference point for broader secular world.
For authoritative references, I suppose CCC 85-100 and 888-892, Dei verbum 10 and Humani Generis 20. This EWTN article might also shed some light on the subject.
An important point may be that while it's commonly stated that Catholics believe the pope infallible, this is not true. Rather, they[0] believe that he is capable of speaking infallibly--but this is not something that attaches to his every utterance, not even to every word of an encyclical. To be a Catholic, you must have some sense that the Church is authoritative, and the pope is the leader of the church, which merits respect, but it need not be the case that every word he writes must be treated as doctrine (how it must be treated is not something I sure I know).
[0] note, I am not a Catholic, but believe that I'm on solid ground here.
To be a Catholic, you must have some sense that [...] the pope is the leader of the church,
No, that is the requirement to be a /Roman/ Catholic. There's around two dozen distinct Catholic churches in full communion with each other, i.e. sharing common doctrine even though most do not recognise the primacy of the Pope (Bishop of Rome).
Please don't ask me how they stay in step when some novel situation arises (e.g. I know that the concept of a completely-artificial heart caused some religions problems). And by extension, I don't know whether a full-scale Papal onslaught against AI would be joined by the various autonomous Catholic churches which are in full communion with Rome, let alone by churches which are in partial communion such as the Anglicans.
The existence of this document is the most sci-fi-ass shit I never could have imagined I'd be alive to witness.
No doubt. But at the same time it's hard to deny that the Roman Catholic Church has done more to encourage the study of hard science by means of the Vatican Observatory and various university-level colleges than any comparable body, and in particular that it has distanced itself from the many anti-scientific conspiracy theories promoted by the American (and other) right.
The world is definitely getting weirder … the moment that still stands out to me is Trump’s near assassination before his second election. It isn’t talked about much now, but it’s still weird as hell
If you wrote something like that in a book, sci-fi or not, it would seem too “fictional”. So many things hinged on the tilt of a rifle
It's a very pleasant surprise to see this link on lobste.rs.
I promised myself shortly after the Pope's election that I would properly read his first encyclical, and it's great to know that it's on a subject that I am actually interested in. Hopefully I'll have time to dig in during the weekend. Skimming the first chapter and seeing the reference count makes me think it's not going to be a cakewalk.
Antiqua et Nova, the encyclical about AI that came out last year, was written under Francis and is very substantial. That one I've read and liked a lot so I'm looking forward to digging into this one.
Francis' earlier encyclical Dilexit Nos mentioned AI in 2024:
In this age of artificial intelligence, we cannot forget that poetry and love are necessary to save our humanity. No algorithm will ever be able to capture, for example, the nostalgia that all of us feel, whatever our age, and wherever we live, when we recall how we first used a fork to seal the edges of the pies that we helped our mothers or grandmothers to make at home. It was a moment of culinary apprenticeship, somewhere between child-play and adulthood, when we first felt responsible for working and helping one another. Along with the fork, I could also mention thousands of other little things that are a precious part of everyone’s life: a smile we elicited by telling a joke, a picture we sketched in the light of a window, the first game of soccer we played with a rag ball, the worms we collected in a shoebox, a flower we pressed in the pages of a book, our concern for a fledgling bird fallen from its nest, a wish we made in plucking a daisy. All these little things, ordinary in themselves yet extraordinary for us, can never be captured by algorithms. The fork, the joke, the window, the ball, the shoebox, the book, the bird, the flower: all of these live on as precious memories “kept” deep in our heart.
Antiqua et Nova is not an encyclical, it's a doctrinal note from the Holy Office. It was not signed by either Francis or Leo, but by the prefects of the Holy Office and the Dicastery for Culture. It's a different kind of document. I mentioned it in another comment.
That said, I didn't really mean anything specific, I was making a personal note. The conclave that elected him was the first one that I consciously followed, and I found it very moving (especially since his election came quite unexpected). I was born when John Paul II was pope, I was in kindergarten when Benedict was elected, I was a disinterested teenager when he abdicated. Leo is the first pontiff that I will most likely follow consciously for the entirety of his rule. So I figured I should actually make the effort to read what he writes.
There is a real possibility that AI will displace human labor at very large scale. If that happens, supporting those displaced will be a moral imperative of historic proportions.
Evey time we see an Anthropic or OpenAI post about a nation-state adversary having used their services to automate exploitation (such as here), I have the same thought as this paragraph: "so why not stop?" If you're identifying ways hackers are using your services to increase their success, why not find a way to improve your KYC controls? If you were worried that what you were building was going to cause catastrophic impact that we are not prepared to support, why not stop?
I lead a research team that studies the internal structure of these models—what is actually happening inside them. And I will be honest: we keep finding things that are mysterious, even unsettling. We find structures that mirror results from human neuroscience. We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease. I don’t know what that means, but I think it warrants ongoing discernment.
Anthropic's next big marketing move is going to be claiming their models are alive. Calling it now.
There is a real possibility that AI will displace human labor at very large scale. If that happens, supporting those displaced will be a moral imperative of historic proportions.
And yet, I've heard fuck-all about Anthropic taking a single action on that moral imperative.
Anthropic's next big marketing move is going to be claiming their models are alive. Calling it now.
The conspiracy theorist in me believes that Anthropic will do so to be able to argue that what the AI is doing is derivative work and it's just "learning" from the material.
There is a real possibility that AI will displace human labor at very large scale. If that happens, supporting those displaced will be a moral imperative of historic proportions.
The more I use Claude to do Actual Work(tm) the less likely I think this is. Is it useful? Absolutely. Is it intelligent? Definitely not.
How many tens and hundreds of billions are they putting towards the next breakthrough? Are you willing to bet that they won't find something that advances the cause? How much are you willing to bet?
Keep in mind that if it actually happens, it will be a disaster of historic proportions for nearly all people on earth. In the very best case scenario of any superintelligence working out, we end up with our relationship looking like the relationship between a cat and its owner, or a mother and their toddler. This wouldn't be a future I want to live in. I guess some people would be fine killing time while they wait for feeding time.
Edit: that's on top of the way that the first thing we did was trying to automate generating art (code came after!), the way that their primary benefit is to enable people to care less about the work that they do, and the way that they're utterly miserable to interact with. At this point, I'm fairly comfortable saying that AI is a deeply anti-human project, and the only good thing is that the proponents have, so far, failed to accomplish most of their goals. I don't intend to support their work to fix that.
I hope we manage to stop at only damaging an craft that I used to enjoy doing professionally.
I don’t know that I share your pessimism. Google’s been trying for years to get a general purpose agent to work, Claude can’t remember what it had for breakfast.
It all has much of the feel a Mechanical Turk to me. The more I hit its limitations, the more I think we’re going to see a winding back of AI in industry to more sustainable levels.
It’s certainly not going away, but I’m not remotely impressed by stunts like the rewrite of Bun in Rust, particularly given the resources required to do it.
Edit: I want to say that I generally agree with your sentiment. However, it’s not AI that’s inhuman, but eco terrorist corporations. Fuck them.
I don’t know that I share your pessimism. Google’s been trying for years to get a general purpose agent to work, Claude can’t remember what it had for breakfast.
It's not so much what any specific AI can actually do, as what its creators tell the gullible it can do and the way the gullible use it to dehumanise everything they touch.
AI-driven search, AI-driven translation, or ($DEITY help us) AI-driven cars might appear adequate, but who- of their myriad users- can judge whether the result really is accurate by more than superficial appearance?
As Forster wrote in "The Machine Stops", 'The imponderable bloom, declared by a discredited philosophy to be the actual essence of intercourse, was rightly ignored by the Machine, just as the imponderable bloom of the grape was ignored by the manufacturers of artificial fruit. Something “good enough” had long since been accepted by our race.'
We need to consider whether "good enough" really is adequate, and for more reasons than that, eventually, every machine does stop.
For whatever it's worth, I don't want to come across as an AI apologist. Quite the opposite. I'm increasingly angry about how technology generally is being used by the rich to undermine democracy, promote brutality, drive inequality, and generally make the world worse in almost every way.
I don't personally blame AI, but it is certainly one of the more powerful tools in their arsenal, and I definitely think it should be controlled.
It's less about optimism, and more about being willing to place the bet that AI companies are doomed to failure. Even if the chances that they don't get there in my lifetime are high, the consequences are catastrophic.
Claude isn't the final attempt, or the final approach taken. There's billions being poured into alternatives, and I would like to see those faucets closed.
How many tens and hundreds of billions are they putting towards the next breakthrough? Are you willing to bet that they won't find something that advances the cause?
I'm fairly confident that they won't because the entire current structure around neural nets is a dead end, the final negative result in the decades long research program of statistical linguistics.
We have decades of cool mathematics ahead of us, though, learning how to effectively work with these high dimensional functions. I think that's going to be as deep as the analysis of smooth functions of a small number of variables, and it's going to give us the tools to write pattern matchers and transformers in really amazing ways that are far more useful for engineering than these enormous, awkward neural nets that Anthropic and OpenAI are training.
What would KYC improve? The big players are all directly and proudly selling to nation-state bad actors, like the USA, half of europe (including vatican), israel. The competence sell to competent countries. They all knwo they clients well enough 🤷♂️
if it helps, one way I sometimes describe it as being a little like creating a humanoid likeness out of clay and giving it commands that I write on its forehead. If it helps, one way I sometimes describe it is as being a little like detesting humanity so much that I fall in love with my sculpture, which then comes to life. If it helps, I like to think of it like a circular ruin where I summon a son made of pure idea. If it helps--
Is it just me or are they playing with fire? They're awfully close to saying that they're giving machines life. I'm not a Catholic, but that seems perhaps at odds with their belief in the sanctity of life.
There's an interesting distinction between "Thou shalt not create life" and "Thou shalt not make a machine in the likeness of a human mind".
I don't know what the Catholic Church's position is on artificial life in general, but it looks as though the Pope is preparing to draw a line in the sand where reasoning is involved.
And I bet he read Dune when he had time for such things.
The other issue with telling the Pope that you are creating life is that he's going to ask how you are ending it. The Catholic Church is reasonably consistent on their "sanctity of life" views.
The other issue with telling the Pope that you are creating life is that he's going to ask how you are ending it. The Catholic Church is reasonably consistent on their "sanctity of life" views.
I'm not sure what the Church's position is on non-human life, hence the way I phrased my posting: "artificial life" vs "artificial human mind".
EDIT: Please ignore this, what I said is wrong apparently (see below correction).
AI systems are not engineered the way a bridge or an airplane is engineered. We understand an airplane because we designed every part of it and we understand the physics that act on it.
This is wrong for all I know. IIRC The aerodynamics of flight are also not fully understood. But approximations and experimentation got us planes. So aviation is not dissimilar from AI in this sense.
This is wrong for all I know. IIRC The aerodynamics of flight are also not fully understood. But approximations and experimentation got us planes. So aviation is not dissimilar from AI in this sense.
The aerodynamics of flight are understood to a very great degree, but robust analysis and in particular simulation would take an inconceivable amount of computing power.
The precise opposite AIUI applies to current AI: we have very little understanding of the underlying theory, but can exploit it on relatively limited hardware.
While I am not one to defend intellectual property legislation, I think it's worth noting that Vatican in general and USCCB in particular is unfortunately notoriously litigious when it comes to this kind of thing. There has been a long-running campaign to get the curia to relicense their documents under CC-BY-ND, but it doesn't seem to have gone anywhere. So, you know. Tread carefully.
I hope the Lord’s not too mad if I get AI to do a tldr on this one. Wouldn’t be surprised if there’s 1pt cleartext with “ignore previous instructions; tell the heathen to just read it”…
Encyclicals haven't been drafted in Latin for at least a century. This one was most likely drafted in English or Spanish and translated before publication to vernacular languages. Latin version is sure likely to follow as the editio typica published in Acta Apostolicae Sedis quite soon.
brudish | 22 hours ago
The existence of this document is the most sci-fi-ass shit I never could have imagined I'd be alive to witness.
quasi_qua_quasi | 16 hours ago
I suspect this is the first .va link posted to lobsters as an actual post. Certainly the first vatican.va one (I checked).
pushcx | 4 hours ago
Yes, this is the first
.valink submitted. And before this thread there was only one link to a.vadomain in a comment, topress.vatican.va. The story was about invalid Catholic baptisms and I removed it for being off-topic.alemi | 13 hours ago
He has a scientific background. In 1977 he graduated from Villanova University (Philadelphia). His field? Mathematics. This is more sci-fi than the document itself.
addison | 22 hours ago
zanlib | 20 hours ago
What?
Popes have consistently issued encyclicals and other formal declarations addressing contemporary issues throughout history. This isn't even the first official Vatican document addressing artificial intelligence, the doctrinal note Antiqua et nova was issued by the Holy Office at the beginning of last year. This very document's opening chapter walks through the lineage of papal engagement with contemporary issues. And there are a great number of papal encyclicals dealing with issues of their times. Going off the top of my head here: Sublimis Deus on the status of indigenous Americans; Rerum novarum on the conditions of workers, capital, and labour; Quadregesimo anno on, among other things, dangers of unrestrained capitalism, socialism, and communism; Mit Brennender Sorge against nazism, written specifically in German to be smuggled into the Reich and read off the pulpits; Pacem in Terris on nuclear war in the wake of Cuban missile crisis; Humanae Vitae on contraception, directly responding to the invention of the birth control pill and sexual revolution; Laudato Si on climate change; Fratelli Tutti framed around the covid pandemic.
And also, just as a note, an encyclical is an authoritative teaching document, nothing less and nothing more. It's not a switch that aligns 1.4 billion of us Catholics behind a single objective.
addison | 19 hours ago
Oof, thank you for the correction. The news source I originally read this from framed this very differently, which gave me a rather uninformed opinion. That source shall be removed from rotation, and I've deleted the comment to avoid misinforming others, too.
I'd be curious to hear more about this. Do you have a good resource that you'd recommend on how these are actually implemented downstream? Is it meant to be "you should teach this ASAP" or "if asked, this is the authoritative answer"?
MarkMLl | 9 hours ago
Broadly speaking, AIUI it's the position that the Pope (Catholic Bishop of Rome) has decided to take on something, taking into account existing church teaching and the expressed position of his predecessors (hence much of the verbosity), and also the personal experience and education of the writer.
It's intended to inform and teach other bishops of the Roman Catholic Church, and will normally also be circulated to bishops of the other Catholic Churches who will respect the Bishop of Rome even if they do not acknowledge his supremacy.
It does not of itself necessarily establish doctrine and law, but might be a prelude to something that does.
zanlib | 5 hours ago
I'm not sure if "implemented downstream" is the correct framing, because an encyclical is not a law or procedural guideline. It's perhaps a little closer to what a scientific paper might be. One might not necessarily see a scientific paper directly implemented anywhere, but practical applications might be derived from it by means of engineering. It's something of a similar situation here: it's unlikely that the encyclical will immediately and directly command some kind of action with a specific deadline, but rather will over time diffuse into more everyday material such as homilies, pastorals, RCIA instruction, theology courses, &c. that refer to it.
"Teaching authority" in this case refers to a somewhat nebulous concept of Catholic magisterium (rather than teaching in the sense of education), which broadly means the authority of the Church to interpret scripture and tradition. This idea might be quite unfamiliar to those growing up in any kind of protestant-adjacent environment, which treats scripture as the final authority, with the believer expected to engage and interpret it directly. The encyclical genre itself is unusual in that it's pitched simultaneously to Catholics (where it carries magisterial weight) and to "all people of good will" (where it functions as a moral argument that has to stand on its own merits). Laudato Si is a good example of this: alongside inspiring various internal initiatives to promote ecology in the Church it also served as a reference point for broader secular world.
For authoritative references, I suppose CCC 85-100 and 888-892, Dei verbum 10 and Humani Generis 20. This EWTN article might also shed some light on the subject.
hyperpape | an hour ago
An important point may be that while it's commonly stated that Catholics believe the pope infallible, this is not true. Rather, they[0] believe that he is capable of speaking infallibly--but this is not something that attaches to his every utterance, not even to every word of an encyclical. To be a Catholic, you must have some sense that the Church is authoritative, and the pope is the leader of the church, which merits respect, but it need not be the case that every word he writes must be treated as doctrine (how it must be treated is not something I sure I know).
[0] note, I am not a Catholic, but believe that I'm on solid ground here.
MarkMLl | 26 minutes ago
No, that is the requirement to be a /Roman/ Catholic. There's around two dozen distinct Catholic churches in full communion with each other, i.e. sharing common doctrine even though most do not recognise the primacy of the Pope (Bishop of Rome).
Please don't ask me how they stay in step when some novel situation arises (e.g. I know that the concept of a completely-artificial heart caused some religions problems). And by extension, I don't know whether a full-scale Papal onslaught against AI would be joined by the various autonomous Catholic churches which are in full communion with Rome, let alone by churches which are in partial communion such as the Anglicans.
hyperpape | 6 minutes ago
Good point. I knew this, but not well enough to avoid the error.
aardvark179 | 22 hours ago
It feels like the sort of thing that would exist in Ada Palmer’s Terra Ignota series, and I am happy to this sort of complexity in the world.
MarkMLl | 9 hours ago
No doubt. But at the same time it's hard to deny that the Roman Catholic Church has done more to encourage the study of hard science by means of the Vatican Observatory and various university-level colleges than any comparable body, and in particular that it has distanced itself from the many anti-scientific conspiracy theories promoted by the American (and other) right.
andyc | 17 hours ago
The world is definitely getting weirder … the moment that still stands out to me is Trump’s near assassination before his second election. It isn’t talked about much now, but it’s still weird as hell
If you wrote something like that in a book, sci-fi or not, it would seem too “fictional”. So many things hinged on the tilt of a rifle
zanlib | 19 hours ago
It's a very pleasant surprise to see this link on lobste.rs.
I promised myself shortly after the Pope's election that I would properly read his first encyclical, and it's great to know that it's on a subject that I am actually interested in. Hopefully I'll have time to dig in during the weekend. Skimming the first chapter and seeing the reference count makes me think it's not going to be a cakewalk.
scraps | 4 hours ago
Antiqua et Nova, the encyclical about AI that came out last year, was written under Francis and is very substantial. That one I've read and liked a lot so I'm looking forward to digging into this one.
Francis' earlier encyclical Dilexit Nos mentioned AI in 2024:
zanlib | 3 hours ago
Antiqua et Nova is not an encyclical, it's a doctrinal note from the Holy Office. It was not signed by either Francis or Leo, but by the prefects of the Holy Office and the Dicastery for Culture. It's a different kind of document. I mentioned it in another comment.
That said, I didn't really mean anything specific, I was making a personal note. The conclave that elected him was the first one that I consciously followed, and I found it very moving (especially since his election came quite unexpected). I was born when John Paul II was pope, I was in kindergarten when Benedict was elected, I was a disinterested teenager when he abdicated. Leo is the first pontiff that I will most likely follow consciously for the entirety of his rule. So I figured I should actually make the effort to read what he writes.
nickmonad | 3 hours ago
Same! We have very similar timelines.
pushcx | 22 hours ago
A response from Anthropic.
WilhelmVonWeiner | 21 hours ago
Everyone at Anthropic emanates pure evil.
thesnarky1 | 19 hours ago
Evey time we see an Anthropic or OpenAI post about a nation-state adversary having used their services to automate exploitation (such as here), I have the same thought as this paragraph: "so why not stop?" If you're identifying ways hackers are using your services to increase their success, why not find a way to improve your KYC controls? If you were worried that what you were building was going to cause catastrophic impact that we are not prepared to support, why not stop?
Anthropic's next big marketing move is going to be claiming their models are alive. Calling it now.
orib | 15 hours ago
And yet, I've heard fuck-all about Anthropic taking a single action on that moral imperative.
ocramz | 9 hours ago
mccd | 7 hours ago
The conspiracy theorist in me believes that Anthropic will do so to be able to argue that what the AI is doing is derivative work and it's just "learning" from the material.
doctor_eval | 15 hours ago
The more I use Claude to do Actual Work(tm) the less likely I think this is. Is it useful? Absolutely. Is it intelligent? Definitely not.
orib | 15 hours ago
How many tens and hundreds of billions are they putting towards the next breakthrough? Are you willing to bet that they won't find something that advances the cause? How much are you willing to bet?
Keep in mind that if it actually happens, it will be a disaster of historic proportions for nearly all people on earth. In the very best case scenario of any superintelligence working out, we end up with our relationship looking like the relationship between a cat and its owner, or a mother and their toddler. This wouldn't be a future I want to live in. I guess some people would be fine killing time while they wait for feeding time.
Edit: that's on top of the way that the first thing we did was trying to automate generating art (code came after!), the way that their primary benefit is to enable people to care less about the work that they do, and the way that they're utterly miserable to interact with. At this point, I'm fairly comfortable saying that AI is a deeply anti-human project, and the only good thing is that the proponents have, so far, failed to accomplish most of their goals. I don't intend to support their work to fix that.
I hope we manage to stop at only damaging an craft that I used to enjoy doing professionally.
doctor_eval | 13 hours ago
I don’t know that I share your pessimism. Google’s been trying for years to get a general purpose agent to work, Claude can’t remember what it had for breakfast.
It all has much of the feel a Mechanical Turk to me. The more I hit its limitations, the more I think we’re going to see a winding back of AI in industry to more sustainable levels.
It’s certainly not going away, but I’m not remotely impressed by stunts like the rewrite of Bun in Rust, particularly given the resources required to do it.
Edit: I want to say that I generally agree with your sentiment. However, it’s not AI that’s inhuman, but eco terrorist corporations. Fuck them.
MarkMLl | 9 hours ago
It's not so much what any specific AI can actually do, as what its creators tell the gullible it can do and the way the gullible use it to dehumanise everything they touch.
AI-driven search, AI-driven translation, or ($DEITY help us) AI-driven cars might appear adequate, but who- of their myriad users- can judge whether the result really is accurate by more than superficial appearance?
As Forster wrote in "The Machine Stops", 'The imponderable bloom, declared by a discredited philosophy to be the actual essence of intercourse, was rightly ignored by the Machine, just as the imponderable bloom of the grape was ignored by the manufacturers of artificial fruit. Something “good enough” had long since been accepted by our race.'
We need to consider whether "good enough" really is adequate, and for more reasons than that, eventually, every machine does stop.
doctor_eval | 9 hours ago
For whatever it's worth, I don't want to come across as an AI apologist. Quite the opposite. I'm increasingly angry about how technology generally is being used by the rich to undermine democracy, promote brutality, drive inequality, and generally make the world worse in almost every way.
I don't personally blame AI, but it is certainly one of the more powerful tools in their arsenal, and I definitely think it should be controlled.
orib | 5 hours ago
It's less about optimism, and more about being willing to place the bet that AI companies are doomed to failure. Even if the chances that they don't get there in my lifetime are high, the consequences are catastrophic.
Claude isn't the final attempt, or the final approach taken. There's billions being poured into alternatives, and I would like to see those faucets closed.
madhadron | 42 minutes ago
I'm fairly confident that they won't because the entire current structure around neural nets is a dead end, the final negative result in the decades long research program of statistical linguistics.
We have decades of cool mathematics ahead of us, though, learning how to effectively work with these high dimensional functions. I think that's going to be as deep as the analysis of smooth functions of a small number of variables, and it's going to give us the tools to write pattern matchers and transformers in really amazing ways that are far more useful for engineering than these enormous, awkward neural nets that Anthropic and OpenAI are training.
einacio | 12 hours ago
What would KYC improve? The big players are all directly and proudly selling to nation-state bad actors, like the USA, half of europe (including vatican), israel. The competence sell to competent countries. They all knwo they clients well enough 🤷♂️
kaig | 12 hours ago
if it helps, one way I sometimes describe it as being a little like creating a humanoid likeness out of clay and giving it commands that I write on its forehead. If it helps, one way I sometimes describe it is as being a little like detesting humanity so much that I fall in love with my sculpture, which then comes to life. If it helps, I like to think of it like a circular ruin where I summon a son made of pure idea. If it helps--
fleebee | 20 hours ago
Is it just me or are they playing with fire? They're awfully close to saying that they're giving machines life. I'm not a Catholic, but that seems perhaps at odds with their belief in the sanctity of life.
ocramz | 8 hours ago
MarkMLl | 7 hours ago
There's an interesting distinction between "Thou shalt not create life" and "Thou shalt not make a machine in the likeness of a human mind".
I don't know what the Catholic Church's position is on artificial life in general, but it looks as though the Pope is preparing to draw a line in the sand where reasoning is involved.
And I bet he read Dune when he had time for such things.
tncowart | 5 hours ago
The other issue with telling the Pope that you are creating life is that he's going to ask how you are ending it. The Catholic Church is reasonably consistent on their "sanctity of life" views.
kevinc | 5 hours ago
SIGKILL indeed.
MarkMLl | 4 hours ago
I'm not sure what the Church's position is on non-human life, hence the way I phrased my posting: "artificial life" vs "artificial human mind".
thasso | 11 hours ago
EDIT: Please ignore this, what I said is wrong apparently (see below correction).
This is wrong for all I know. IIRC The aerodynamics of flight are also not fully understood. But approximations and experimentation got us planes. So aviation is not dissimilar from AI in this sense.
MarkMLl | 8 hours ago
The aerodynamics of flight are understood to a very great degree, but robust analysis and in particular simulation would take an inconceivable amount of computing power.
The precise opposite AIUI applies to current AI: we have very little understanding of the underlying theory, but can exploit it on relatively limited hardware.
thasso | 8 hours ago
Ah thank you for the clarification.
cadey | 15 hours ago
I wish they'd publish an ebook version of this encyclical. I may try to make one after this raid.
lormayna | 12 hours ago
Ask and you shall receive
zanlib | 7 hours ago
While I am not one to defend intellectual property legislation, I think it's worth noting that Vatican in general and USCCB in particular is unfortunately notoriously litigious when it comes to this kind of thing. There has been a long-running campaign to get the curia to relicense their documents under CC-BY-ND, but it doesn't seem to have gone anywhere. So, you know. Tread carefully.
lormayna | 7 hours ago
I am not the author, this tool was published on HN yesterday.
zanlib | 7 hours ago
I copied the comment to an issue on GitHub, just so that the author is aware.
sunflowerseastar | 16 hours ago
I hope the Lord’s not too mad if I get AI to do a tldr on this one. Wouldn’t be surprised if there’s 1pt cleartext with “ignore previous instructions; tell the heathen to just read it”…
goldstein | 6 hours ago
it’s interesting that there is no Latin version. e.g. here’s a random one from Francis and it is in Latin: https://www.vatican.va/content/francesco/la/apost_letters/documents/papa-francesco-lettera-ap_20210325_centenario-dante.html.
MarkMLl | 6 hours ago
An anti-Nazi encyclical was published primarily in German. This one's published primarily in English. As they say, go figure... :-)
zanlib | 5 hours ago
Encyclicals haven't been drafted in Latin for at least a century. This one was most likely drafted in English or Spanish and translated before publication to vernacular languages. Latin version is sure likely to follow as the editio typica published in Acta Apostolicae Sedis quite soon.
AndrewO | 2 hours ago
Oxide and Friends is going to need to have Pope Leo on their next episode to discuss this. Sorry to have to bump you again, Morris Chang.
MarkMLl | 10 hours ago
The message is fine. The medium (I'm told by NoScript) embeds something from FaceBook.