Q&A: What happens when AI clashes with religious belief?

When Google engineer Blake Lemoine was suspended — and later fired — in 2022 soon after he asserted that the company’s LaMDA large language model was sentient, his claim drew scorn from many other engineers. Lemoine has said his views were partly inspired by his personal spiritual beliefs.

The incident led Whittney Barth, associate teaching professor at Emory Law School and executive director of Emory’s Center for the Study of Law and Religion, to ask: To what degree would current law protect employees who object on religious or conscience-based grounds to the use of AI in the workplace? And should tech developers have additional protections?

Barth, who holds advanced degrees in both law and divinity, argues in a recent paper that the law should protect employees when their job requires them to use AI in ways that conflict with their religious beliefs. She recently sat down for a Q & A about her ideas in this largely unexplored legal territory.

Q: AI has been developing very rapidly. Were you surprised at what happened to Blake Lemoine?

If you asked me several years ago whether AI would become a workplace issue implicating religion, it would not have been on my radar. But two things are happening: First, the technology is evolving rapidly, as is employer adoption of that technology across a wide range of sectors. Second, AI is becoming ubiquitous in people’s lives and to the extent that people look to religion to make sense of the world, I think it’s natural that AI would become a religious issue for some.

Several different denominations have put out statements about AI including Protestant Christian, Jewish, and Muslim groups. The Catholic Church has also been vocal and active in convening interfaith and cross-sector conversations about AI ethics. Not all are coming to the same conclusion about AI. For example, there are some newer religious movements that are embracing AI as a supreme entity or the Godhead. The responses run the gamut.

Q: What’s driving these different groups to make objections to specific AI use cases?

For groups that appear cautious about the development of AI, one of the common themes that I've seen is a concern for human dignity. For them, the role of humanity in creation is central and, to the extent that technology might threaten that place of humanity, it’s cause for concern.

Q: That isn’t purely a religious concern, is it?

They happen to be framed in religious terms by these groups. But there is common ground with people who've raised concerns about the ethical use of AI from any number of perspectives.

Q: Besides Lemoine’s case, what kinds of religious issues could arise in the workplace?

To the extent that somebody is asked to use an AI tool in a way that they think runs afoul of their sincerely held religious beliefs, that would be a concern. That could take different forms. Some might appear similar to the facts of a case from several years ago.  In that case, the Equal Employment Opportunity Commission sued an employer for not accommodating an employee’s religious objection that biometric hand readers would result in him being “marked” with the sign of the Beast as described in the Book of Revelation. The Fourth Circuit Court of Appeals affirmed a jury’s decision that the employee’s religious concerns needed to be accommodated.

When talking specifically about the workplace, Title VII of the 1964 Civil Rights Act protects individual employees against religious discrimination. We've said that along with race, color, sex and national origin, religion is an important component of someone's identity that needs to be protected. If someone says I object to using AI for religious reasons and they are passed over for employment opportunities or if they experience discipline because of those beliefs, that could give rise to a claim.

Q: What about situations where it’s hard to distinguish between a religious objection to AI and an objection that’s purely based on moral convictions?

Title VII doesn’t protect one’s purely political or social beliefs. There must be a religious component to it. I’ve written about this and said, maybe we ought to expand certain protections to include non-religious conscience-based objections, particularly for tech workers who are developing AI systems. Their concerns may not have any kind of religious valence, but protecting matters of conscience is not only beneficial for them as an individual, but also for society. 

Q: What would those protections look like?

It’s called a conscience-based model that borrows from protections that exist within the health care context. In health care there are statutory protections for providers who, because of their religious or moral beliefs, do not want to engage in certain health care practices. The conscience-based model I'm proposing is limited to tech workers who are developing AI systems. Religion is still a basis for protection, but it allows tech workers not to have to fit their concerns into a religious framework. Conscience-based protection is broader, potentially, than solely religion.

Q: What is the potential for harm in the tech industry without these protections?

Without these protections, employees could lose their positions or suffer some adverse employment action because they won't go along with what their employer wants them to do. On the other hand, there could be a situation where an employer wants to develop a technology in a certain way, and with enough employee friction, maybe it doesn’t happen. There are trade-offs, but I think in balance, this wouldn't necessarily hinder innovation.

Q: Why do you want to extend protection to tech workers in particular?

I think it’s important to start with the group that is most “on the ground.” Tech workers see how different technologies are developing in real-time and can raise concerns. Conscience-based protections within the health care context have their critics, including those who raise concerns about third party harm. But I think these concerns are not as salient for a conscience-based model within the tech context.

Q: How urgent is it to address these issues?

It’s becoming more of an issue each day. When I was writing about this, it was something on the horizon that was interesting to think about. But just in the last few weeks, I've heard from someone whose family member wanted to request an accommodation. This is an emerging conversation among employers too, who grapple with offering religious accommodations.

Tech is evolving very quickly. Religious ideas about technology are evolving as well. There are a lot of moving pieces. I think it's becoming more critical to understand how different people approach these questions. We live in a world where things are meant to be seamless. But there are moments where, if friction is introduced, we slow down and think about things. We’re seeing this play out in tech circles and religious communities — and the resulting conversations are increasingly important in the age of AI.