Pope Leo XIV’s AI Encyclical + Anthropic’s Three Questions. ARD #84

Michael Parekh AI: Reset to Zero

18 hours ago

16 MIN READ

Today’s theme is focused around Anthropic’s reactive call for three discernments to the Pope and the Vatican. It was an invitation for Anthropic to present alongside Pope Leo XIV, who was presenting a major AI Encyclical — Magnifica Humanitas.

In a 42,000+ word document, Pope Leo called out at least six key things: disarming AI (protecting from nuclear-energy-type downside risks); calling AI ‘fundamentally non-Human’; AI in warfare and automated lethality; labor, inequality, and “digital slavery”; truth, democracy, and youth protection; and environmental impact.

Of the six points, the second one — calling AI fundamentally non-Human — is the one I would literally underline in terms of my support. It’s important to separate AI technology as a technology and not to anthropomorphize it. Most humans will be susceptible to considering AI human because that’s our tendency — we do that with pets and animals, so there’s no reason we wouldn’t do that with technology. Especially when the technology companies are going out of their way to make these things as likable, friendly, and reaffirming as possible.

The invitation to Anthropic — a call to one of the leading AI technologists of our time — was an opportunity to address these points on both sides of the spectrum of possibilities, and their probabilities. It was the opportunity for an important dialog with the Church and its 1.4+ billion followers, to help sort out the possible major issues around this AI Tech Wave — constantly stirred up in the broader conversation around AI for the past few years, in terms of the doomerism and the acceleration movement around AI in general.

Let’s go through the three Questions cited by Chris Olah — Anthropic’s eighth founder, and Interpretability Research Lead (his job is to lead the teams that figure out how the AI models work in the deep recesses of their matrix maps and tables) — and discuss them along with what the Pope lays out in his AI Encyclical. Followed by my Take.

(1) Anthropic’s First Question — Duty to the Global Poor

Chris Olah’s first Question to the Pope and the Vatican, on the duty to the global poor:

“The first is our duty to the global poor. There is a real possibility that AI will displace human labor at a very large scale. If that happens, supporting those displaced will be a moral imperative of historic proportions. This task will be difficult enough, but I worry most dialogue misses an even harder challenge. AI development is concentrated in a handful of wealthy nations. How can we ensure the gains of AI are shared globally? We do not have a mechanism for this. It is an unsolved problem, and it is the kind of problem the Church has historically refused to let the world ignore.”

(2) Anthropic’s Second Question — Moral Imagination and Ambition Regarding Human Flourishing

Chris Olah’s second Question:

“The second is the need for moral imagination and ambition regarding human flourishing. If AI models are going to be widespread, what does it look like for humans, families, and the world to flourish? Today, parents are already worried about their children’s minds; individuals about the future of their work. These are not questions a lab can answer but they are questions traditions like yours have carried for millennia, and we need you to keep carrying them into this new moment in history.”

(3) Anthropic’s Third Question — Discernment on the Nature of AI Models

Chris Olah’s third Question:

“The third is the need for discernment on the nature of AI models. I am a scientist. 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.”

MP TAKE — The Glass-Half-Full Counter

Each of the points by the Church, and the questions presented by Anthropic, are well laid out — but directionally they reflect the Glass Half Empty overall in all of the arguments and concerns around AI. Without any probabilities laid out to the possibilities.

The invitation to Anthropic was an opportunity to address these points on both sides of the spectrum of possibilities in my mind. Each of the six points and more brought up by the AI Encyclical were well laid out. But Anthropic’s response and questions, in my view, only represented one side of the spectrum of possibilities — with implications of high probabilities attached to each. And that’s one of my core issues with this. There should have been a Glass Half Full element to all of this.

As I’ve written a lot in these pages and talked about on this podcast, technology for hundreds of years has invariably resulted in far greater global prosperity despite the challenges of the old way of doing things versus the new. Technology, as per Jevons Paradox and other economic things to study, has vastly expanded market opportunities by orders of magnitude over time. So the world’s poor have generally benefited massively because of technology over the last couple of hundred years. And AI is no different in the context of past tech waves.

Yes, they may be accelerating, happening faster, more things at the same time. Those are all elements to be managed by better governance, clear-eyed thinking around this, and clear laying out of the probabilities around some of these possibilities. And other countries, by the way, in my view — China and others — are really thinking through this on a bottom-up basis very differently than what’s going on in the US and these kinds of debates. That is very important to be able to figure this out.

And while we don’t understand all the technical intricacies of the AI models, the key point I would highlight is that they are still mostly mirrors to humans. They’re based on modeling our neuroscience — what little we understand about it (we maybe understand 10 to 15 percent of how the human brain works, if that). Then we tend to fuel them, train them on data that humans have generated in our languages. And then we run probabilities on all of that. Chris Olah, who is the chief scientist who figures this out and knows this — this is just looking at humans and their existence through their language in the mirror. And that’s what the models are spitting out. Ultimately it’s literally looking into a massive hall of mirrors.

So to then say that this is a whole new thing that we have to have these concerns around, I think is a little bit of an exaggeration. And I wish there was a Glass Half Full part of this discussion. To fear the models is to fear ourselves as humans. And so far we’ve done alright thus far coming onto the first quarter of the 21st century. So balancing out Anthropic’s questions with the Glass Half Full part of the questions would be timely to do as well.

Gadget AI — American Airlines Picks SpaceX’s Starlink for In-Flight WiFi on 500+ Planes

Yet another airline, Amerocan this time, picks SpaceX Starlink for their in-flight WiFi broadband. We’re going from narrowband dial-up speeds on airplanes to broadband. American joins United and a bunch of other airlines to pick Starlink. Obviously very timely given Elon just filed his mega AI IPO at a valuation of $1.7+ trillion.

CNBC outlined the deal this week — “American Airlines picks SpaceX Starlink for in-flight WiFi”. American Airlines’s own announcement — “American to install Starlink — the fastest Wi-Fi in the sky” — confirms the partnership.

For longtime readers — the Starlink realities and context is in AI-RTZ #1007, “Elon Musk SpaceX/xAI IPO Starlink”. And Amazon, Globalstar, and Apple also in the satellite-to-device mix is in AI-RTZ #1057, “Amazon Leo Buys Globalstar Satellites”. Amazon with its Leo constellation just bought GlobalStar for $11+ billion, along with Apple — who’s a major shareholder.

MP Take: This is the beginning salvo in an array of higher-speed broadband on commercial flights. SpaceX Starlink in the pole position, but others coming on strong. Hundreds of planes will take a while to refit with new dishes. In the next two or three years we’re going to finally have broadband speeds like we enjoy in our homes and offices over the last couple of decades. And it’s about time.

Questions

Q1 — What does MP view as the other major positive of this announcement?

Answer: It adds to other airlines like United and others who are also adding broadband WiFi access to their planes.

What I’m most excited about — besides having the broadband speeds when we travel — is that for now, most of the airlines are not planning on charging extra dollars like they do for narrowband WiFi today. I just had a flight where I had to pay $8 to United for just a three-hour flight, or $49 a month for their WiFi. It looks like right now competitively, most of these companies — even though these are oligopolies — are basically saying: if you sign up to our Airline Loyalty Club for free, we’ll give you the broadband access for free. But I suspect that may only be a short-term promotional thing.

Q2 — What concerns MP most about higher-speed broadband on planes?

Answer: The oligopolistic nature of the airline market, combined with the higher cost of these Starlink connections, means the airlines may decide to charge for these connections.

The airlines, because they’re oligopolies, can’t help themselves and they basically end up charging for the higher speeds. And Starlink right now, by the way, is already raising prices on the Pentagon for their high-speed WiFi. So I wouldn’t be surprised if this also happens once Starlink/SpaceX is public — Starlink raises prices on the airlines too. These are already at rates of tens of thousands of dollars a month for clusters of airplanes. The pricing is complex. So that may give the airlines an excuse to step back from the free pricing for broadband. We’ll see what it ends up costing if they go back to their usual ways.

MP Take: The 2-to-3-year refit timeline plus Starlink/Amazon/Apple satellite-to-device competition is the structural counterweight to the oligopoly-pricing risk. Watch the Amazon Leo + Globalstar + Apple thread closely — if that constellation lights up at competitive prices, the airlines’ free-broadband-with-loyalty model has a chance of sticking.

Source Reading — For the Full Context

For the full context, see the canonical sources:

The Vatican’s AI Encyclical Magnifica Humanitas itself — at vatican.va.
The New York Times — “Main Takeaways from Pope Leo’s Encyclical on AI” — the mainstream summary of the six key points.
Anthropic’s own write-up — “Chris Olah on Pope Leo’s Encyclical” — the company’s official framing of the visit.
The Washington Post — “Anthropic aligns with Vatican over White House as Pope Leo stokes AI fears” — the political-positioning frame.
Wired — “Why the Vatican invited Anthropic” — the why-Anthropic-specifically frame.

For longtime readers — five backcat references frame the broader AI-philosophical-discourse context:

AI-RTZ #879, “Waiting for AI-God-Like AGI” — the canonical MP frame on AI-as-deity discourse.
AI-RTZ #1092, “Regular Folks in the US Want to Slow Down on AI” — the mainstream-sentiment counterpoint to the Vatican’s slowdown framing.
AI-RTZ — Marc Andreessen’s Techno-Optimist Manifesto — the canonical AI-optimist counter-manifesto.
AI-RTZ — AI on the Shoulders of Giants (OTSOG) — the Newton-via-Merton frame on AI as standing on prior generations’ work.
AI-RTZ #1082, “Limiting AI with Human Intelligence in the Mirror” — the canonical MP frame on AI-as-mirror-of-humans.

Shorts Clips from today

Clip 1 — AI: A Mirror to Humans

Watch on YouTube Shorts

While we don’t understand all the technical intricacies of AI models, they are still mostly mirrors to humans — based on modeling our neuroscience, trained on data humans have generated in our languages. Chris Olah, the chief scientist who figures this out, knows this — this is just looking at humans and their existence through their language in the mirror. A massive hall of mirrors.

MP Take: To fear the models is to fear ourselves as humans. And so far we’ve done alright thus far coming onto the first quarter of the 21st century. Other countries — China and others — are really thinking through this on a bottom-up basis very differently than what’s going on in the US debates. The Glass Half Full counter to Anthropic’s framing is the missing balance.

Clip 2 — AI’s Duty to the Global Poor

Watch on YouTube Shorts

Anthropic’s First Question to the Pope and the Vatican: there is a real possibility AI will displace human labor at very large scale. AI development is concentrated in a handful of wealthy nations. How can we ensure the gains of AI are shared globally? Unsolved problem, says Anthropic — the kind the Church has historically refused to let the world ignore.

MP Take: So far so good — but all the other tech waves of the past had no problem with general dissemination around the world for the benefit of global populations. PCs, smartphones, social media, digital internet — all extraordinarily global. Each country does their own versions of it, but it’s been diffused. We need to remember the historical context of other technology waves.

Clip 3 — AI: Next Tech Wave for Global Prosperity

Watch on YouTube Shorts

Technology for hundreds of years has invariably resulted in far greater global prosperity despite the challenges of old versus new. Technology, as per Jevons Paradox, has vastly expanded market opportunities by orders of magnitude. The world’s poor have generally benefited massively because of technology over the last couple hundred years. AI is no different in the context of past tech waves.

MP Take: Yes, AI may be accelerating — happening faster, more things at the same time. Those are elements to be managed by better governance, clear-eyed thinking, and clear laying out of probabilities around possibilities. China and others are thinking through this on a bottom-up basis very differently. Balancing out Anthropic’s questions with the Glass Half Full part of the spectrum would be timely.

AI Ramblings Daily on AI-RTZ is here to think through AI and reset. Together.

Tomorrow — ARD 85 on AI-RTZ #1100.

Thanks for joining us, AI Curious Folk. Stay tuned.

NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.