Mi3: Every day, millions of people ask Alexa or Siri for help—often using words like ‘please’ and ‘thank you.’ But what if we’re getting too comfortable with these digital ‘friends’—and it’s changing us in ways we don’t even realize? Today, we’re exploring the question with Rebecca Geiselmann, a doctoral researcher from Mi3. Rebecca is a cognitive neuroscientist and has authored a piece called “Interacting with Agents Without a Mind: The Case for Artificial Agents.” She is also active in several science communication activities with citizens and students. But let’s start with her key question: are people attributing human traits to AI, even when it doesn’t truly have a mind? To start, can you explain why this topic is so relevant today?
Rebecca: Absolutely. We interact with AI in various forms—virtual assistants, chatbots, robots—and there’s a growing tendency for people to treat these systems as if they have human-like qualities. This isn’t just about how sophisticated the AI is; it’s also about how we, as humans, perceive and interact with these systems. The main question we explored is whether people attribute human traits, like agency and the capacity to have experiences, to AI, and what implications that might have.
Mi3: That’s certainly something I do! So, what does your research reveal about how people humanize AI?
Rebecca: What we see is that when AI starts to look or act more like us—think of a humanoid robot, for example—we’re more inclined to see it as something that can act with intention, almost like it has its own agency. But here’s where it gets interesting: even though we might start to think of AI as having some level of intentionality, we’re still pretty reluctant to believe that it can truly have thoughts, or consciousness, or have desires like we do. So, while we might begin to treat AI as if it can act on its own, we don’t quite see it as having a mind in the same way we see other people.
Mi3: So, the brain plays a big part in how we perceive AI, but it doesn’t completely buy into the idea that AI has a mind?
Rebecca: Exactly! Our brains are quite sophisticated in distinguishing between a machine and a human, even if the machine acts or looks somewhat human-like. For instance, certain brain regions light up that are usually active when we’re observing and understanding actions—like the mirror neuron system. This system is all about recognizing and representing actions, so when we see a robot moving or doing something, our brain processes it in a way that’s somewhat similar to how it would if a human were doing the same thing. This helps us to attribute “agency” to the AI, meaning we see it as something capable of intentional actions.
But the brain areas that deal with understanding others’ experiences, like the temporoparietal junction (TPJ) and the medial prefrontal cortex (mPFC), don’t get as activated when we’re interacting with AI as they do with humans. We might engage with AI ‘as if’ it has agency, but we don’t fully extend that to believing it has genuine experiences.
Mi3: What do you mean by “as if” then?
Rebecca: The idea that we interact with AI ‘as if’ is a significant philosophical question. It raises deep questions about how much of our interaction is based on a suspension of disbelief and how much is influenced by our cognitive processes. Do we realize that we act ‘as if’ AI has a mind to make our interactions smoother and more intuitive? This would mean that deep down, we know it’s not the same as interacting with another person.
Mi3: Can you give an example of how this plays out in everyday interactions with AI?
Rebecca: Sure! Take virtual assistants like Siri or Alexa. People might ask them to play music or set reminders, and they often use polite language—saying “please” and “thank you.” This kind of interaction shows a level of agency attribution, where users treat the AI as a service provider that can understand and respond to requests. But if you were to ask people if they believe Siri or Alexa can actually feel frustrated when they can’t fulfill a command, most would probably say no. They recognize that these systems can perform tasks, but they don’t think of them as having subjective experiences. It’s a kind of selective anthropomorphism.
Mi3: That’s an interesting question! When we interact with AI, especially something like a robot, our brain gets pretty involved. So, does that mean we see AI as having a mind like a human?
Rebecca: Not exactly. While our brains might treat the AI as an intentional agent—something that can act and make decisions—we don’t fully attribute a mind to it, especially when it comes to experiencing emotions or thoughts.
Mi3: That’s really fascinating. It seems like the human-like design of AI plays a big role in how we interact with it. What are the implications of this for our future interactions with AI, especially as these systems become more integrated into our lives?
Rebecca: Design is crucial. When AI is designed to resemble human appearance or behavior, people are more likely to interpret its actions as intentional, compared to when it appears more mechanical or artificial, which can have both positive and negative effects. On the positive side, this can make AI more intuitive and accessible, encouraging people to use it more effectively. However, there’s also a risk that people might overestimate the capabilities of AI, assuming it has a deeper understanding or empathy than it actually does. This could lead to misplaced trust or unrealistic expectations, especially in critical areas like healthcare or customer service.
Mi3: And you believe there’s a social dimension to this, right?
Rebecca: Absolutely, that’s one of the key insights from our paper. We’re not just influenced by our own perceptions of AI, but also by how others around us interact with it. If you see someone treating an AI as if it’s human, you’re more likely to follow suit. This is where the research gets really interesting—other work at MI3 is exploring these collective dimensions too. For example, we’re looking at how social norms might evolve as AI starts to learn from how we treat them, or how our ways of talking might change. A lot of research tends to focus on individual attitudes in isolation, but what we’re seeing is really more of a collective shift in how we relate to AI.
Mi3: Given these insights, what’s next for your research? How do you plan to explore this topic further?
Rebecca: First, we want to better understand the neural mechanisms behind how people attribute these human traits to AI. What’s happening in the brain when someone starts to see a machine as more than just a tool – something with agency but not quite a full mind? And with the rise of social media and online interactions, there’s the question of whether these digital platforms are encouraging an “illusion of social connection” where people feel like they’re interacting with others when they’re really engaging with algorithms. This could have huge implications for how we understand and manage our relationships with technology moving forward.
Mi3: It sounds like there’s a lot more to uncover! Thank you for sharing these insights!
Rebecca: Thank you! It’s an exciting field indeed, and it is great to talk to you.
Check out the original study and Rebecca Geiselmann’s profile.
References:
Geiselmann, R., Tsourgianni, A., Deroy, O., & Harris, L.T. (2023). Interacting with agents without a mind: the case for artificial agents. Current Opinion in Behavioral Sciences, 51, 101282, doi.org/10.1016/j.cobeha.2023.101282