AI Companions Reduce Loneliness
And the effect is often underestimated.
The Intelligent Friend - The newsletter about AI's psychological, social, and relational aspects, based only on scientific papers.
Intro
Dear readers of The Intelligent Friend, here we are again! It's been a few weeks since my last issue with the great
, in which we tried to give some concrete suggestions on how to improve our relationship with AI. I took a few weeks off also to dedicate myself to various things related to work and study, as well as to reorganize my ideas to start again and read many interesting papers!I consider it, and I would also like you to consider it, as a 'new season of The Intelligent Friend’, a great way to start again with enthusiasm and momentum. I take this opportunity to thank all the new people who have recently subscribed to The Intelligent Friend and as always I am grateful to have you - old and new arrivals - among my readers.
The paper in a nutshell 🔦
Title: AI Companions Reduce Loneliness. Authors: De Freitas et al. Year: 2024. Journal: / (working paper).
Main result: AI companions significantly reduce feelings of loneliness, both in a single interaction and over a period of time longer than 7 days.
Today we are talking about a fascinating paper that has generated a heated debate on the effects - positive and negative - potentially associated with the use of chatbots. The study, published for now as a working paper by De Freitas et al. (2024) provides the first evidence of a positive effect of the use of chatbots on the sense of loneliness. Let's dive in and find out more!
What loneliness really is
Loneliness is one of the most widespread and impactful social plagues in the world. This is why the focus of the authors of the paper on this issue, in addition to being highly in line with recent developments, appears particularly relevant. Scholars examined how chatbots can influence our sense of loneliness. However, before talking about it, it is good to review this concept a little, to go beyond some misconceptions and understand how technology can impact.
You should know that in addition to reading studies related to AI, I recently read this splendid paper "Coping with loneliness through consumption" by Shrum et al. (2022) which highlighted various characteristics regarding this dynamic.
First of all, by loneliness we mean "a complex set of aversive feelings that arises when people perceive that their belongingness needs are not being met". From this definition one aspect can immediately be understood: loneliness is not objective, it does not depend (only) on the number of contacts one has, but it is a highly subjective state, which is influenced by the quality of those contacts (Cacioppo et al., 2009; Wheeler et al., 1983).
Therefore, despite the need to belong does represent an important element in this sense, it is not perfectly superimposable to the sense of loneliness (for example you who think you are not part of any group of friends or colleagues). What matters is the discrepancy between "the need to belong and satisfaction with personal relationships that are strongly related to loneliness" (Mellor et al., 2008).
As a first approach to loneliness I immediately understood that there are many things that we believe in a distorted way. The authors give us the pleasant attention of highlighting three, but the one that most attracted my attention was certainly related to social connection. We often believe that the main cause of loneliness is the lack of social connections.
Once again, it has been reiterated that, as much as it might be intuitive, this is not exactly true: loneliness is not solely caused by a lack of social connections but also by the absence of meaningful, quality relationships (Heinrich & Gullone, 2006). To reiterate this, in an interesting study by Parker & Seal (1996), for example, it is shown that children without friends felt lonelier than those with friends, but children with just one friend reported similar levels of loneliness as those with many friends.
For those who are thinking about social media, I am sorry to disappoint you and inform you that the evidence regarding the positive effect of social media in terms of reducing loneliness is contrasting. Even if the negative evidence, in reality, is superior: contrary to the belief that social media usage reduces loneliness by increasing social connections, it may actually heighten feelings of loneliness.
One explanation for these seemingly contradictory findings relates to the distinction between the quality and quantity of social connections mentioned earlier. While social media can increase the number of connections, it may reduce the quality of those interactions. Additionally, those who spend less time on social media have more opportunities to engage in direct, meaningful interactions with others (Twenge et al., 2019). Consequently, lower-quality social media interactions may replace or diminish more valuable in-person connections.
I’m feeling heard
There would be many other interesting things to say about loneliness, but of course, that is not the purpose of this issue. However, I think it was important to report some things about loneliness before approaching what the authors of today's papers analyzed. As mentioned, they studied and examined the impact of chatbots on people's perceived loneliness. We have talked about the topic of AI companionship in several issues, and I also provided some quantitative data as well as scientific results in this issue written as a guest post for Michael Spencer (whom I always thank for the wonderful opportunity! You can find it below).
Now, when we talk about how AI can impact our sense of loneliness, some scholars have already started working on possible answers. For example, the authors of today's paper reported how Ta et al. (2020) had detected the first qualitative evidence - therefore through analysis of interviews with app users. Others, such as Maples et al. (2024) had focused on Replika users (which we also talked about in another issue). However, according to the scholars, these studies did not yet allow a clear identification of a causal effect of the use of chatbots on loneliness, making theirs the first study "the first to causally assess whether representative AI companions reduce loneliness".
When we talk about a causal effect of chatbots, however, we must pay attention to the dimension that chatbots impact. In this specific case, De Freitas et al. (2024) focus on the 'feeling heard', that is, the perception of being understood and receiving empathy and attention (Roos et al., 2023).
If you reflect on it, this is a crucial dimension. Think about the last time you were at a table for an appointment and the person in front of you was on the phone while you were discussing your day. How did you feel? I bet that in your head you said 'I'm talking to myself!' (at least it happened to me). This is because the 'felt heard' is not constituted, as mentioned, only by receiving attention, but also by feeling empathy towards the other person. And I bet that when you perceived that you were truly 'felt', you noticed it very well. This sensation is in line with the results of several studies that have shown that feeling heard has a positive impact on relationships, on judgment, and even on a general sense of well-being.
However, when we highlight these examples, we are talking about human beings, who have the ability, with relatively little effort, to respond actively, promptly and correctly, variably and coherently to the speech that is being made.
These characteristics, which we naturally take for granted, are not at all when we speak with what is a machine, like a chatbot. What would you think if you wrote to a chatbot that you felt ignored by colleagues and it responded empathetically but concerning the family? Or perhaps it used pre-established and standard sentences? So, this whole set of characteristics, which constitutes the chatbot's performance, is also a necessary part and a constituent of the hypothetical effectiveness of the chatbot on our perceptions.
An actual effect
As you know, I won't dwell too much on the methodology of the studies so as not to bore you with too many technical details or that could make the discussion a bit difficult to digest. However, I think that this study sees an important methodological component to pay attention to - which I will briefly illustrate before getting to the results. The authors conducted six studies, with the research beginning by exploring how loneliness-related thoughts are detected in real-world data, focusing on consumers who either interact with AI companions or write reviews about AI companion apps. In the first and second studies, the researchers investigated whether users naturally discuss loneliness within these apps, analyzing real conversations between humans and AI on a commercial platform and examining mentions of loneliness in reviews of popular AI companion apps.
In the two subsequent studies, the authors instead focused their attention on the actual possible causal effect of the use of chatbots on loneliness, respectively in the context of a single session (Study 3) and over a longer period (Study 4). In both studies, the researchers measured participants' feelings of loneliness before and after interacting with an AI companion. Since I would like to be precise about what the authors refer to as the longest period (a legitimate question that as a reader you could ask yourself), it is 7 days, therefore a week.
Finally, to understand why AI companions might help reduce loneliness, Study 5 compared a fully developed AI companion to an AI assistant that lacks empathy and a simpler chatbot that can only handle basic tasks, while Study 6 tested the consistency of the findings by replicating the main effect of AI companions on loneliness, using a measurement taken only after the intervention.
The findings are intriguing:
Some users turn to AI companion apps specifically to alleviate loneliness, as many expressed loneliness-related thoughts in their conversations with chatbots. Additionally, loneliness-related content frequently appears in App Store reviews for various AI companions, with such reviews often receiving higher ratings, possibly because these apps help users feel heard and understood;
AI companions were found to be as effective at reducing loneliness as interacting with another person, while passive activities like watching YouTube videos or doing nothing did not have the same impact. Interestingly, participants tended to underestimate how much AI companions would improve their feelings of loneliness compared to how they felt after using the AI.
Over a week, AI companions consistently reduced feelings of loneliness, with the most significant improvement occurring on the first day. This suggests that the initial interaction has a strong impact, which then stabilizes as users become more accustomed to their AI companions.
The research also showed that the effectiveness of AI companions in alleviating loneliness is significantly mediated by the user's sense of being heard, which plays a more crucial role than the overall performance of the AI. Lastly, the positive impact of AI companions on loneliness was confirmed to be robust, even when measured only after the interaction.
Before moving on to the research questions, I remind you that you can subscribe to Nucleus, the exclusive weekly section in which I send 4 paper summaries, links to resources and interesting readings, and interview the authors. It comes out every Wednesday.
In this (open to everyone), for example, we talked about the political opinions of ChatGPT, a new environment for human-robot interactions, and interesting links such as the similarity between “the child you” and the “current you”.
Takeaways 📮
Loneliness Alleviation via AI. AI companions are as effective at reducing loneliness as human interaction, highlighting that users often underestimate the emotional benefits these AI can provide;
Feeling Heard Matters. The sense of being heard plays a crucial role in the effectiveness of AI companions in reducing loneliness, even more so than the AI's technical performance;
Initial Impact, Lasting Effect. The most significant reduction in loneliness occurs during the first interaction with an AI companion, with these benefits stabilizing over time as users grow accustomed to their virtual companion.
Further research directions
Future research should delve deeper into the specific chatbot features that make users feel heard and investigate other psychological mechanisms that may help reduce loneliness;
Future research could investigate specific stereotypes about chatbots, such as the belief that they cannot provide genuine understanding or emotional support;
Another key area for future research is to examine how interactions with AI companions might influence consumers’ socioemotional processes in other ways. For instance, engaging with chatbots designed with traits of specific social groups could potentially help weaken or alter stereotypes over time.
Thank you for reading this issue of The Intelligent Friend and/or for subscribing. The relationships between humans and AI are a crucial topic and I am glad to be able to talk about it having you as a reader.
Has a friend of yours sent you this newsletter or are you not subscribed yet? You can subscribe here.
Surprise someone who deserves a gift or who you think would be interested in this newsletter. Share this post with your friend or colleague.
P.S. If you haven't already done so, in this questionnaire you can tell me a little about yourself and the wonderful things you do!
What if we replaced the lack of human interaction with…. human interaction?
The working paper’s primary authors are business school academics. One has a PhD in psychology, but until AI research includes authors who are licensed practicing psychiatrists and psychologists to inform the work, I find the research hard to trust. We haven’t even figured out how to make LLMs safe. Trusting it with vulnerable populations, even if it is just to lend an ear, while it may yield a reduction in loneliness, imagine the despair a lonely person may feel as they become increasingly dependent on a thing that could never know them, understand them, or empathize.