Beyond What AI Can Do: Religion, Ethics, and Human Connection

Jonathan Beasley: Artificial intelligence has become the subject of intense public debate. Some people are excited by what it can do. Others are concerned about its effect on jobs, communities, and the growing power of the companies developing it.

These are not only questions about technology. They are also questions about ethics: who gets to shape the future of AI, how it should be used, and how we can benefit from it without losing something important about the way people learn, think, and relate to one another.

I’m Jonathan Beasley, and this is the Harvard Religion Beat.

Today, I’m joined by David Lamberth, Professor of Philosophy and Theology at Harvard Divinity School. His scholarship explores religious experience, questions about the mind, and how people make meaning.

We spoke about the concerns driving the public debate over AI, what these systems can and cannot do, and how people and organizations can make more thoughtful decisions about using them.

Jonathan Beasley: Your scholarship has explored religious experience and questions about the mind and how people make meaning. How did those interests lead you to questions of artificial intelligence and ethics?

David Lamberth: My work has long connected with ethics, working on the pragmatic tradition. The ethical is always pretty closely within reach, and I’ve had that set of interests across theological materials and materials in the study of religion. But the interest in artificial intelligence really popped up on my screen in connection with my long-standing thinking about philosophy of mind and psychology and the interaction of questions there.

But it popped up on the screen, really, in November of 2022, when ChatGPT 3.5 dropped. And like everyone else who took notice of that, it was this massive leap forward in terms of LLMs at least being able to function in a way that started to look a lot closer to what we would expect, meaning something like the Turing test or that of a conversation partner who you can’t tell whether they are or aren’t human.

One of my professors when I was a grad student here in the PhD program was Hilary Putnam, who is a really an important figure in thinking about artificial intelligence. Putnam came out with the theory of functionalism, which really anchors a lot of the thinking that went into the work, building up to where artificial intelligence is now.

Interestingly, Putnam decided that he was wrong about that. And so he wrote a book in 1988, refuting his own view. And that book came out the year after I started in the PhD program, and he was one of my two advisors all the way through my dissertation. So these questions have always been on the margin for me, and they just came back to the center as things started to cook really.

Jonathan Beasley: And you started to notice it as well. When people talk now about AI ethics, what issues are you finding that tend to receive the most attention? And then what are the important questions do you think are being overlooked?

David Lamberth: In the university context, obviously, one of the major questions is just around pedagogy and student education and whether people are doing their own work and how they’re interfacing with AI and whether the work we’re reading when we’re evaluating students is actually theirs. And there are broader questions about that in the scholarly domain.

I think when we think about the larger public, if we’re just thinking about the broadest public, I think people are more aware of the infrastructure questions. So data centers, for example, what kind of impact those are having, where those are showing up. So people jump immediately from the question of the actual AI capabilities to the infrastructure that’s being built and invested in to support those.

I think people have fewer tuned in kinds of questions about the ethics of what these leaps that are being made in each technological iteration imply. And that’s one of the places where we’re really interested in helping people think concretely about what’s going on.

So very often, the public thinks in terms of threat or harm or risk where the flip side of that about thinking about what does this technology afford, what are the possibilities, what’s the future going to look like as this technology proceeds to be a part of how we do things. Those are the kind of questions. Ethics often gets framed up as a negative emphasis, as a focus on what you should not do. But obviously, thinking about the good and about human flourishing is really the key purpose in many respects for a lot of ethical discourse.

Jonathan Beasley: Hmm interesting. AI is constantly in the news now, reading stories, some terrifying, some maybe not so terrifying, but it’s still in the news. And nearly everyone seems to have a strong opinion about it. We’re seeing local opposition to data centers growing distrust of technology leaders who are often in the news and then concerns about AI agents acting unpredictably. What do these different reactions tell us about the public’s underlying ethical concerns?

David Lamberth: I think the public is in a position right now where we have a widespread distrust of anyone who’s exercising power. And it’s really power more than authority. Obviously, the leaders of this relatively small number of technology companies are wielding an immense amount of power on the development of these systems and have philosophies that vary fairly significantly from one another. But also, they probably vary wildly from what the public might agree would be a good or proper way to move ahead.

And our current politics is really not engaging these questions. So I’ve talked with people before. And on the question of data centers, I was talking with a student the other day who lives in a place where there’s a big data center that was built in a neighborhood that was economically depressed and was talking about how we have to be able to do something about that. And I said, well, the one part of politics that is still working relatively well across the country is local politics.

And so the data center question is, in many respects, one that you can see the possibility for engaging politics from the ground up to the middle level. So you see in Massachusetts, you see communities that don’t want to have data centers and change their zoning and put in levels of review.

Very much like electrical generation or telephone distribution, where the switches go in the old school copper system, it really is a distributed network phenomenon. But it raises the equity question of who gets saddled with the downside of where these things go.

Jonathan Beasley: I want to talk about your recent Harvard Divinity Bulletin essay. You argue that discussions of AI can oversimplify how people think and make meaning in particular. What parts of human understanding and experience tend to get left out of those conversations?

David Lamberth: Most of what we’ve been talking about in the AI domain over the last three years has had to do with large language models, these chatbots, which in the old days prior to 2022, chatbots were these annoying things you would bump into on the internet that would be doing really haphazard things and be unimpressive.

And the scale shift was when OpenAI released ChatGPT 3.5, and it was really a pretty functional thing. And then we’ve had iterations from OpenAI and Anthropic putting out Claude and iterating on that. You’ve got xAI’s Grok product. You’ve got now Google Gemini in the domain. And we’ve also got companies from China that are producing DeepSeek and Kimi, for example, that are models that are really very close to this.

All these companies are producing these really effective language processing models. Now, we very often present ourselves to one another through language. And so these models being able to master predictively what human language roughly sounds like and then interact with these massive stores of data that are being organized mean that these models have, in certain ways, achieved much greater intelligence than any one of us could command.

It’s interesting. When I was in college, a professor could make their whole career off of being the master of the 20 works of one particular thinker, because you really didn’t have the Control F function on a PDF database of what’s there. But these LLM systems have access to really everything that is currently in the digitally published sphere of humankind.

And they have their finger on that button. You and I could never read that much, much less command and recall it in that way. So it simulates something that looks extraordinarily like what the apex of human intelligence might be.

But what it’s leaving out is all of the rest of that human part of the intelligence equation. And actually, I don’t necessarily think that intelligence is the only human capacity that we really value by any stretch of the imagination. We’re embodied beings. We are flesh and blood. We feel. We actually feel while we think. We’re feeling how our thoughts are progressing.

And William James famously argued in one of his essays that the sentiment of rationality, the idea that a thought is a logical and good thought is actually something that we feel as a lack of impediment to our thinking. So it’s kind of like flow.

So consciousness and our sense of self and our sense of continuity over time, our sense of where our thoughts are going and what they’re choosing and not choosing, all of that’s inflected by this affective dimension, which is completely missing in a large language model, which is functioning with a predictive algorithm of which words go next in the context of the subject matter that I’m accessing information on at the moment.

So it’s this great synthetic power of conceptual content, but it lacks any of the embodied or affected dimension that’s really constitutive of human experience, whether individually or certainly in relation to other people.

Jonathan Beasley: You’re listening to the Harvard Religion Beat from Harvard Divinity School. My guest today is HDS professor David Lamberth. We’ve been talking about why AI ethics involves more than preventing harm. It also requires us to think about human flourishing, who holds power over these technologies, and what AI may leave out about the way people think, feel, or relate to one another.

Professor Lamberth explores these questions further in his Harvard Divinity Bulletin essay, “Meaning Making, Bodies, and AI.” You’ll find a link in the show notes.

If you’re enjoying this conversation, please give this show a follow on your preferred podcast platform. For more from Harvard Divinity School, including new podcast episodes, articles, and events, follow HDS on social media. You’ll find those links in the show notes as well.

Now, back to my conversation with David Lamberth.

Jonathan Beasley: Professor Lamberth, What are some of the ethical risks that do arise when AI is treated primarily as a technical tool but without enough attention to how it affects those human relationships, communities, and the way we’re making meaning?

David Lamberth: Treating it merely as a technical tool can often lead us to just completely ignore the impact that it has on individual’s sense of themselves, on their sense of what they’re doing, whether it’s learning in the context of a school setting.

It’s those human engagements that really help us through those moments of despair. And so insofar as you lean on the technical capacity, that’s a problem. If we swing back to the educational environment, for students, AI often is thought of as just a temptation. But of course, it’s a very powerful tool for doing excellent research, for expanding your thinking, for helping you understand things better.

So there are all these positive possibilities. And at the same time, if we outsource central parts of our thinking and capability developing processes to an LLM, then what we see is what colleagues at MIT have referred to as de-skilling.

And there’s a recent study that came out in China, recent being a few weeks ago. But students who were using AI to prepare for a standard set of exams actually did better on the exams in the short term, but then when they were retested at a longer term, actually had less retention and less capability.

So you can see how—and that could be true in the business world as well, where if you outsource too much of what you’re doing, you may lose a handle on how to understand who your market is, or in the domain of religious leadership, if you outsource interacting with your members of your church community or your religious community, you may lose that ability to really have this developed, textured sense of what’s really going on in the community.

So there’s this big possibility, which is counterbalanced by this set of risks, the harms of which may range from world destruction, but they may come all the way back down to you, Jonathan, losing your capability to do things that you 20 years ago were really good at.

Jonathan Beasley: We talked about some of the few leaders, really, who are at the front of all this. Ultimately, as we continue to move through all of the development and the latest iterations, who should be responsible for ensuring that AI is developed and used ethically? Should it be the companies, the leaders, governments, individual users, a combination? Who should it be?

David Lamberth: We’re at an interesting inflection point where the industry has been allowed to develop pretty much freely without any large intervention. Obviously, there’s been a major investment support. So there’s a way in which these companies are able to spend just massive amounts of money in these developments, both in the infrastructure and developing the software,

Because the future looking folks who want to invest in that from the venture capital worlds and elsewhere and eventually in IPOs that OpenAI and Anthropic are planning, we’re just going to see massive amounts of money come in there.

Thinking about society at large, in the United States, at least in Europe, we function as democracies. We think that the people overall, through their representatives, should be able to weigh in on questions of import like this, questions which have to do with interstate commerce and infrastructure and education and daily life, all these things are ordinary topics that are within scope for state and local and federal governments to deal with.

We haven’t yet seen any action at scale to really engage with these questions. And they’re really live questions about whether the industry is moving so fast that even if there were regulatory frameworks that the regulators wouldn’t be able to keep up. That doesn’t mean that we shouldn’t be doing that. I mean, I fundamentally believe that the Democratic basis of our social structures really should be the basis for how we think about these momentous questions and companies like this, just because they have a lot of money shouldn’t necessarily be able to put whatever product they want into play or introduce risks that are really untenable and not social goods in any respect into the mix.

But politics is complicated. And of course, AI development is not only going on in the United States, it’s also going on in China where the structure of economic decisions and the governing authorities and the relationship to the broader governance populace is working on a very different model.

Jonathan Beasley: For people making decisions about whether and how to use AI in their workplaces or organizations, what questions should they be asking? Where should they be? What should they be looking at?

David Lamberth: Well, I think one of the keys is to develop the capacity to step back and really ask the broader questions of what’s the benefit and what’s the potential risk or harm for deploying these tools in this domain. And of course, the questions about whether AI is going to keep the economy moving forward in a productivity increase or whether it’s going to cost U.S. jobs and really cause a collapse in our standard economic form is this really large question. But you can ask those kind of questions on a smaller scale.

What am I trying to achieve here? How in the broader context of the enterprise I’m engaged in, whether it’s private enterprise or a public enterprise. How might we leverage the good and minimize the risk? How do I keep growing the skill set of the workforce that I’m interacting with. How do I do that with as a professor in a university with students?

But how do I also think about using these tools to leverage information and make advances on the research field. Being an AI skeptic and even AI abstinent is probably not the right stance to take. Being skeptical about it is totally fine. But the question of thinking that now, while I’m just going to be abstinent from AI, I’ll just avoid it in all respects, I think you have to think about that as domain specific, because everything we do now is interfacing with this technology, and this is where we’re going to be 10 years from now and 20 years from now is using this in everything we do. The question is, how.

How do we make this a plus in our lives and not take away the parts that are most valuable to us as humans, as individuals, as people in relationship with one another, as members of communities?

Jonathan Beasley: In January 2027, you’ll be co-leading with Richard Geruson HDS’s executive seminar on leading ethically in the age of AI. What do you hope participants will take away from it? And how might they apply what they learn in their various fields?

David Lamberth: We have people within this upcoming session for about three days, 2 and 1/2 days. So it’s a pretty fast-paced plan for the seminar, but we’re hoping to give people the tools to be able to think about this question of goods and harms and risks, and think about them from the perspective of multiple ethical frameworks and also multiple religious frameworks.

One of the interesting things about leading in any kind of organization really, certainly in the United States, but many other places in the world is our workforce, our constituency, our student body is a very diverse group of people.

So how do you think about the impact of what you’re doing, not only on your customer base or who what your work is oriented to but also on your workforce. How do you engage with those different viewpoints? Be aware that they’re there. And how do you also assess the products for how they behave in circumstances that crucial to what you’re doing. So we’re trying to give people multiple lenses to think about ethics from and make those explicit, whether it’s talking about utilitarian or happiness based models or thinking about a rule based approach, or thinking about the common good as a basis, whether and at the same time thinking about how different religious frameworks inflect that.

So if you’re coming from a traditional African framework think about the community as constituting the individual, not as all of us as discrete individuals thinking. So we’re trying to widen people’s frameworks in a way and give them tools for analyzing not only how they’re interacting with the technology but what the technology actually is itself.

Jonathan Beasley: It’s exciting. It seems like a great opportunity.

David Lamberth: We’re looking forward to it. Working with Richard Geruson’s been really terrific. He’s got a background in the tech world as a long time upper management CEO in the tech world, and he’s also a graduate of our Master’s of Religion and Public Life program, where he did a project on the ethics of AI and how to think about that as framework. So he’s really been a terrific partner in this.

Jonathan Beasley: That’s great. Professor Lamberth, thank you for your time.

David Lamberth: Really my pleasure. Thanks, Jonathan.

Jonathan Beasley: The debate over artificial intelligence is often framed as a choice between embracing new technology and resisting it.

But as David Lamberth reminds us, the ethical questions are more complicated. We need to consider who shapes these technologies, who benefits from them, and what happens when systems that process enormous amounts of information are treated as if they understand the world in the same way people do.

We also need to ask what we may be giving up when we rely on AI and whether a tool that can potentially make life or work easier, could also weaken the abilities and relationships that make us human.

David Lamberth is professor of philosophy and theology at Harvard Divinity School. You can read his essay, “Meaning Making, Bodies, and AI,” in the Harvard Divinity Bulletin.

You can also learn more about “Leading Ethically in the Age of AI,” the HDS executive seminar he will co-lead in January 2027. Links are in the show notes.

The Harvard Religion Beat is presented by the Office of Communications at Harvard Divinity School. It is produced and hosted by me, Jonathan Beasley, and edited by Tyler Sprouse.

If you enjoyed this episode, please follow the podcast, leave a rating, and share it with others.

Thanks for listening. Until next time.