Ethical Machines

How far does a data scientist’s responsibility reach? Does it stop at building an accurate model, or does it extend to what the model does once it’s in use? Ben Green, professor at the University of Michigan, former data scientist for the City of Boston, and author of Algorithmic Realism: Data Science Practices to Promote Social Justice, thinks it extends a long way: being a good data scientist means thinking about overall impact. I press him on whether being good at data science and being an ethically good data scientist are the same thing, whether data scientists are in a position to see how their work will play out, what to do when they disagree about what a good outcome is, and how much we should ask of individual data scientists versus their field and the systems they work in.

What happens when students turn to LLMs to learn about history? My guest, Nuno Moniz, Associate Research Professor at the University of Notre Dame, argues this can ultimately lead to mass confusion, which in turn can lead to tragic conflicts. There are at least three sources of that confusion: AI hallucinations, misinformation spreading, and biased interpretations of history getting the upper hand. Exactly how bad this can get and what we’re supposed to do about it isn’t obvious, but Nuno has some suggestions.. Previously aired in season three.

Today I’m talking with Elisha Wiesel, chair of the Elie Wiesel Foundation and former CIO of Goldman Sachs. We talk about why the foundation won’t use AI to recreate his father (Nobel Prizer winner and author of "Night," which recounts his experience in Auschwitz), why he refuses to say what his father would have thought about today’s politics, whether LLMs should show where their ideas come from and whether anyone would care if they did, what the Talmud can teach us about preserving minority opinions and saying “I don’t know,” and what financial models and LLMs get wrong in the same ways.

I roll my eyes at talk of "AI welfare," the idea that we should take seriously the notion that AI has interests or a well-being that we should take seriously. My guests, Geoff Keeling and Winnie Street, both Fellows at the Institute of Philosophy at University of London and co-authors of “Emerging Questions in AI Welfare," don't think that anything like today's AI models are conscious or have a welfare we need to concern ourselves with. But they do want to take seriously questions about whether they could ever have a welfare, what that would take, how morally costly it would be if we failed to attribute well-being to entities that have one and how costly it would be if we attribute well-being to entities that have none. Buckle up - it's a complex and fascinating discussion.

We often think of AI companions as bad, or at least suspect. Giada Pistilli, formerly Principal Ethicist at Hugging Face and now at Mistral, wants to test that assumption, and she's done a lot of research to back up her view. In her new book, The Essential Guide to AI Companions, she looks at people who form emotional relationships with chatbots as partners, as therapists, and as stand-ins for parents who have died. We talk about how people end up in these relationships, why most of them weren't looking for one, whether attachment to a chatbot is a problem in itself, and whether chatbot memory should be off by default.

My guest today is Christian B. Miller, professor of philosophy and author of The Honesty Crisis. Christian's central claim is that technology (AI, social media, etc.) has created conditions where dishonesty is increasingly incentivized and easier to get away with. When you put those two together, you get an honesty crisis.

We discuss, as some examples of growing dishonesty, students using LLMs to write papers and passing off the work as their own, romantic relationships with AI and whether having one could constitute cheating, celebrity and influencer culture's relationship to bullshit and self-deception, and the surprising problem of pastors plagiarizing sermons with AI. While the conversation was about an alleged and plausible crisis, it turned out to be a very fun one.

Season four kicks off with Paula Goldman, author of the new book, “Manage the Machine: How to Harness Human AI Collaboration at Work,” Paula is also Salesforce’s Chief Ethical and Humane Use Officer. We talk about the expertise pipeline problem, why humans should be "at the help" instead of "in the loop," and the ways responsibility gets murky when dealing with teams comprised of people and AI agents.

Here are three paradoxes, according to Virginia Dignum, my guest today: 1. The more capable AI becomes, the more it reveals the richness and complexity of human intelligence. 2. Less bias in AI does not necessarily create more justice. 3. The pursuit of artificial superintelligence may ultimately reveal that humanity's greatest intelligence is collective, not artificial.

We discuss the limits of computation, the dangers of confusing data with reality, why AI ethics often misses deeper social problems, and what it would mean to build technology that genuinely serves human flourishing rather than replacing it. Our conversation is grounded in the book “The AI Paradox”, authored by Virginia, who is a professor in Responsible Artificial Intelligence and the Director of the AI Policy Lab at Umeå University in Sweden.

Jonathan Schaeffer thinks we're building AI the wrong way.

While large language models have produced remarkable results, he argues that hallucinations, bias, and unreliability aren't bugs that can be fixed—they're consequences of the underlying architecture itself. In his view, LLMs are an important stepping stone, but not the path to the kind of AI we can truly trust.

We discuss whether current AI systems are "good enough," automation bias, AI regulation, data centers, environmental costs, and the race toward AGI. We also debate whether society should slow down long enough to put meaningful guardrails in place before deploying increasingly powerful AI systems at scale.

Jonathan Schaeffer is a Professor of Computing Science at the University of Alberta and a pioneer in artificial intelligence research.

My guest, Mona Sloane, author of Predicted: How AI Is Restructuring Social Life, argues that AI has become part of our social infrastructure. Its predictive systems increasingly shape how we work, find information, build relationships, and navigate society.

Mona worries that as prediction becomes embedded in more areas of life, we risk becoming less willing to deliberate, challenge assumptions, and shape our own futures. I push back on whether AI really should be understood as infrastructure and whether predictions made by AI are fundamentally different from the predictions humans have always made.

We also discuss democracy, power, regulation, and what happens when prediction becomes the dominant way of understanding the world.

Book: https://a.co/d/04GwwuFR