The Pope, the Overton Window and the Real Fight Over AI Ethics

ai Leadership Insight series. I'm your host, Mike Vizard. Today, we're joined by Yvette Schmidt, who is CEO of Fusion Collective, and we're having a little chat about AI ethics and what the Pope has to say about all this stuff.

Yvette, welcome to the show. Thanks, Mike. Thanks for having me back.

It's good to be here. Are we finally having a moment about maybe AI ethics that we've been kind of dancing around but not taking too seriously? It's nice that the Pope is talking about this, but I'm starting to see other religious leaders strike a similar tone.

So is something going on here? That's a really good question, and I would say no. If people are thinking that this is going to be the moment that the Pope or any other religious leader is going to create, draft any type of legislation or law, they kind of, you're missing the point.

The thing that I would like to call out here is that the Pope's, I think it's called an encyclical document. I'm just going to call it the Pope's paper. So the Pope's essay, his note to the world, basically is about changing the way that we all think about ethics and AI, and when it comes to humanity as a whole.

So I'm just going to call him Pope Leo, my good friend Leo, because this is something that he's been talking about for a while. And so when people think that all of a sudden his manifesto or his document is going to change the future of AI regulation, I'm going to say yes in a way, but not the way that people expect. So, like I said, the Pope isn't going to write a single line of a statute.

He is not going to do that. And anyone waiting for him to do that is completely missing the point. But what his encyclical document does is that it moves the Overton window.

So Mike, do you know about the Overton window? No, but go ahead. Explain.

Okay. All right. I'm sure most people don't.

So, it's a political term. So the Overton window is a model for understanding what is, and most importantly, what it isn't acceptable in politics, because this is all where it's going to come down to, right? And inside the Overton window, there sits what is considered socially acceptable, right?

Which is like the center of what is policy is supposed to be made of. And outside of that window sits as what is considered radical or unthinkable ideas. And so what's contained in the Overton window shifts over time.

I'm going to use the example here back in the United States, racial segregation in the US, right after Plessy versus Ferguson, the Supreme Court decision in 1896. The decision allowed for separate but equal accommodations of people of color, Black people, and white individuals. To many, at that time in 1896 and for a few decades to follow, that idea of desegregation was far outside of the Overton window.

But on May 17th, 1954, the Plessy v. Ferguson decision was ruled unconstitutional, leading to the desegregation across the United States, right? So half a century from the original decision, many people, white people mostly, thought that desegregation was too radical to be a policy.

So now this is a perfect example of how the Overton window shifts over time. It can shift, it can expand or can contract over time. So people can change their minds sometimes, right?

Events can reshape their views and cultural changes makes ideas trendy or like passe. So when that happens, new policies, once unthinkable, can become adopted. So what the encyclical document does, it hands legislators and regulators and boards the moral cover to say the thing that lobbyists have been spent years making politically expensive to say.

That self-certification isn't accountability and that concentrated, unaccountable AI powers is a problem worth governing. So that's what the Pope's paper does. And just think about his reach.

This document addressed the entire Catholic Church, right? More than a billion people, and beyond them, all the good people of goodwill. So that's the social proof at scale.

No blog post, including mine, will ever touch. So when the diagnosis that we as practitioners have been making for years gets restated from that pulpit, it stops sounding like activism and tree hugging and theoretical speaking, and it starts sounding like consensus. And that changes boardroom conversations.

It changes what a regulator can credibly propose. But for me, it's like this. We're going to watch the unveiling itself for the tell.

So the document states plainly that AI imitates human intelligence. It doesn't feel joy or pain. But guess who was sitting up there in that auditorium with the Pope and a whole bunch of other people was the co-founder of Anthropic, Chris Ola, who's standing beside the Pope saying, oh, his team finds internal, it states that functionality mirror grief and joy.

So the industry wants it both ways. They want to embrace the moral framing in the Cathedral of St. James, but keep shipping out the very thing the framing warns about.

So that tension is the whole reason that influence and enforcement aren't going to be the same word. And round of applause at the Vatican's auditorium changes absolutely nothing if the incentives underneath all of it go untouched. So from my perspective, you don't need Mike.

I don't sell hopium, right? Moral authority isn't enforcement, and we talked about what happened with the New York State Raise Act. So we have to see.

For me, if 40,000 words from my friend Leo the Pope can finally make the boardroom hear what we've been saying all along, I'm going to takeThe Scottie Pippen assist. And while the document sets the weather, whether it rains or not, well, that's going to be really up to us. " And yet, there was a report now, I think it was on NBC, talking about how a lot of the AI companies are saying one thing and lobbying for something else when they go into these various state legislations and various things that people are proposing.

And there's a disconnect, to be kind, between what people are saying and what they might actually be doing. So how do we wrap our arms around what we actually want to have happen? Well, from an industry perspective, from an individual perspective, and then there's the corporate perspective.

So I'm going to sit with the corporate perspective real quick. And it is going to be very hard for organizations to continue to try to skirt the issue of building systems that have irreparable harm on people and think that that's going to last forever. Right?

To think that that's going to happen, without any type of swinging of the pendulum, is very shortsighted. " I think what people, leaders of organizations need to do is embed responsible AI practices into their culture, their workflows, procurement, and decision-making, right? And this doesn't mean that you have to go hire thousands of minions and build some transformation program to bring in some big four consultants to do some transformation program in your organization.

What you need to do is four things, and this literally you can do today in most organizations. One is, okay, you've got AI. If you're not telling people you got AI, you probably got it running somewhere, or someone in your organization is running it shoddily, right?

So one, do the honest absence audit, right? Who's not in the room? Not just who's not, what consequential data is missing?

Who's missing? From the training data, from the design room, from the test population, and then write down that answer even if it hurts a little bit. Then two, move governance out of legal.

So you know what? Ethics that lives only in legal or the post-incident review or lawsuit, that's not ethics, right? Put accountability checkpoints inside the development workflow before the launch, before the lawsuit.

And three, name the power concentration in your stack. Do you know who owns the models and the infrastructure that your company depends on? What are the terms?

And what's your plan when those terms change? That's not a procurement decision right then. That's a long-term strategic decision.

And then make that procurement question with real weight when you negotiate. And then four, demand transparency. This is not like a PDF summary document, right?

And I say a well-formatted PDF is not accountability. Ask for the logic, the data, the populations the system was tested against. And treat proprietary, that's not the black box, right, as the start of a negotiation, not the end of one.

And none of these that I rattle off, Mike, requires a headcount. It just requires deciding that the model in production matters more than that framework slogan on your wall, right? And I believe that the organizations that make that decision now will look brilliant in 18 months, right?

Because the ones that wait will be doing it under the deadline that someone else has set. That's my roundabout way of giving four steps of what people can do today that doesn't really cost any money. It just costs you being willing to be accountable.

So walk through this a little bit because I sometimes wonder, is it the algorithm that's problematic or is it the data that we're exposing to the algorithm? Because I think a lot of the biases and issues that we're exposing when we apply AI to something existed before the AI came along, and it was based in our, whether it was redlining real estate districts that led to education issues and tax-based things. That's not the AI's fault.

That's the data and the system's fault that we had beforehand. So how much of this is something that AI did, or how much is this is AI is exposing things that we already did to ourselves in a way that we don't want to see? So, that's a great question, and here's the thing.

It's just like Chris Olah said it, everyone has said it, and everyone knows this to be true. It's the data is created by us. It learned from our writings, the crazy stuff on the internet, everything that's in your organization, websites, good, bad, or indifferent, right?

It learned from that. So it's not that it's the AI. The technology is neutral, right?

You tell it what to do, it's going to go do it. It has no feeling, even though Chris Olah said it kind of mimics ours. But the framing is this, is thatNobody wants to be responsible for what the outcomes from learning on us results from technology.

That's the bigger issue, right? So it's like there's a difference between actually being accountable and then not. Again, calling for accountability.

One of the press release and the other one has names on findings. And so I want people to understand that it's not the technology, it's that all the isms of humans, it learned from that. So Chris Olah named it, right?

He named the fallout from the most powerful chair in the room of what happens when it's learning from us, all of our isms, all of those things. Then ask the church, a 2,000-year-old institution, to carry the moral water, and he's one of the few people positioned to actually act on all the problems that he's naming, but he doesn't want to be accountable. All the issues that result from learning on all our isms is human-made.

Technology will never be able to fix that, Mike. We have to shift left and go back to what's being a good human. What's being a good neighbor?

Those things will never be fixed, and doing the handoff to the church, like, "Here's the moral water. " We all have conscience, but it's all about decisions and choices that we make. And if you don't want to be accountable, yes, you can make that someone else's problem.

So what ultimately do we need to do to resolve this? " The vendors aren't going to do it. The politicians have shown themselves to be susceptible to, once again, anybody who has a large amount of cash and seems to want to contribute to a campaign, and suddenly there's all kinds of rules and regs that got slid by without anybody understanding what they meant.

So where does this fall? Corporations themselves, they don't always love regulation, and when they do want it, they just want it to be consistent and maybe as loosely defined as possible. So who's going to stand up and say, "We need to do something here"?

We do. And so we always come back to this, Mike, when you and I chit-chat. Okay, so let's go back to what Chris Olah really said during his statement.

" Right? He also said that all the frontier models, including Anthropic, operate inside incentives that can conflict with doing the right thing. That's what he said.

Those are quotes. Go look at it. He said it.

So you're saying, what can we do? What can we do? If we are hoping or waiting for the pope's paper, I'm going to just say that, to make a big splash and big change, we can't wait on that, right?

We'll be waiting till the fifth of Never-uary. But what I want to give you as well as your friends, your families, and everyone that listens to your show, this is what I want us to do. I want us to remember that we own our data, and your data today is more valuable than your vote.

And the companies that are building the systems know that. And most folks running around here living their best life on the internet don't. And so here is how all of us can start acting like we do.

Read the terms of service. I say this all the time. Yes, I know.

I know they're long. I'm not saying read all of it. They want you to scroll.

Yes, they do that on purpose. Why? So you just scroll to the bottom and click the accept so you can make your squirrel video or whatever, or find out if you're a werewolf elf witchy witch, whatever that is, right?

They do it long so you can just hit the button, right? You don't have to read all of it, but find the data sharing section, find the arbitration clause, find the part about what they can do with your information after you delete your account. All those folks that signed up for DeepSeek, yeah, and you deleted your account, well, they still got you.

They didn't read the terms of service. They put it plainly in there. So you're literally signing a contract, so know what the other party is walking away with.

Remember, you own your data, so act like it. So I said it before, every personality quiz on social media is a data collection mechanism. Every free tool is a product where you are the inventory.

So make that trade-off consciously, not by default, right? And then opt out where you can. Most platforms bury the settings deliberately.

I'm telling you, it took me a whole 20 minutes to opt out of a platform, who I will not name, allegedly, to opt out. It takes you 20 minutes, but those 20 minutes are worth it because they bury them in the settings like that, hard, deliberately so you give up, so you won't opt out. But they're still there.

Data brokers, ad tracking, behavioral tracking. It's going to take 20 minutes, and that is what standing up for your own data sovereignty costs you right now. Ask one question before you download anything new.

Like anything. What is this company's business model? What's its business model?

Right. So that's how we can protect ourselves. I do think it's not a matter of if or when, I do believe we're going to get regulation, but a lot of things are going to happen before then.

So the onus, Mike, is on me, you, you telling your friends, me telling my friends, all the listeners that come to your podcasts and your shows, reading your books and your posts. We have to educate the end user because assuming that, oh, the Pope said it, so they're going to do the right thing, or assuming them self-grading their homework by saying that it's safe, assuming that they're going to do the right thing and hold that social contract because since it's out in the world and everybody's using it, it's going to be safe, is faulty. You have to own your data like you own it and move accordingly until the regulations catch up.

All right. So last question, do you think that the reason the Pope's letter is so long is that he was writing it not just for us, but maybe he was hoping the AI would read it, too? Probably.

It is posted on the Vatican's website, which is readable by all AI search engines. So who knows? Who knows?

All right, folks. Well, you heard it here. Words matter even more in the age of AI, so pay attention to what they say and read those fine print clauses really closely because, as Yvette noted, if you can't figure out what the business model, it means you are the product.

Hey, Yvette, thanks for being on the show. Thanks, Mike. All right.

ai Leadership Insight series. You can find this episode and others on our website, and we invite you to check them all out. Until then, we'll see you next time.