Very little research on this topic makes me feel unsettled. However, this latest Harvard Business School (HBS) study has achieved something noteworthy.
AI Companions Reduce Loneliness, proved how much AI companions can reduce loneliness just as effectively as human interaction.
Let me start with a story and explain to you why I am concerned.
Reading this paper took me back to my childhood.
My parents were rarely with me, often away, busy with work or dealing with their own troubles. I grew up with my granny, who didn’t know much but made sure I was dressed warmly and fed well.
I remember clutching my Tamagotchi, that pixelated pet needing constant attention. For a while, it was my best friend.
I responded when it beeped. It needed me to break the egg, grow, and live. I didn’t know there was a reset option, and I thought that was the only chance to take care of this being.
I have a very different life now.
I have a partner with whom I understand each other deeply; we can chat about any topic. I also have friends I can talk to when I’m bored.
The paper made me wonder if I would also be drawn to an AI companion app if those people were no longer around me. Relying on an AI companion could mean entrusting my emotional well-being to a for-profit organization.
This is concerning because a corporation aims to optimize its revenue and market growth. Without strong regulations, there is little protection against how our interactions with AI could be used for monetization.
How do you think your AI friends might prioritize your relationship vs. the company’s profits?
In this article, I will
* Review the HBS research that suggests AI companions can alleviate loneliness.
* Then, uncover the ethical dilemmas and dangers of advertising within AI companionship.
The Digital Quest to Cure Loneliness
Using tech to beat loneliness isn’t new. We are not even talking about AOL, pornhub or social media here. At least there were humans on the other side, well… most of the time.
Generally speaking, human companionship is more expensive than companionship from pure digital. The idea of digital companionship was born in the late 1990s.
A Brief Landscape of Digital Companionship
Tamagotchi, in the late 1990s, was one of the first globally phenomenal digital pets.
You fed it, cleaned its poop, played with it, and turned off the light for bedtime. It had sold over 82 million units worldwide by the late 2010s.
Then, AI companions and chatbots designed explicitly for emotional support emerged in the late 2010s:
* Replika, marketed as an AI friend who’s always there to listen. As of August 2024, Replika has over 30 million users.
* Xiaoice by Microsoft: Over 660 million users, showcasing a massive demand for AI companionship in addressing loneliness.
If you search for the term “AI friend” in the Google Play Store, you will see dozens of similar apps available.
Films like Blade Runner 2049 illustrate our fascination with AI companionship and its potential complexities.
Blade Runner 2049- The “Love” of The Holographic AI
Here’s a clip from the scene in Blade Runner 2049 where K gives Joi a new portable device, allowing her to go anywhere.
I find this scene both tender and bittersweet. Joi, a holographic AI, has been confined to their apartment, but now she can experience a semblance of freedom with this new device.
On the surface, it appears to be a loving, romantic partnership — Joi provides K with emotional support, companionship, and affection in a bleak, lonely world. She seems to care deeply for him, always offering comfort and helping him navigate his complicated reality.
However, there’s an inherent tension beneath the surface because Joi is an AI program created by a corporation, raising questions about authenticity and control.
This raises questions about the authenticity of their relationship. Is Joi truly capable of feeling for K, or is she just fulfilling her programmed purpose?
It’s a poignant dynamic.
Back to reality. This research from HBS and what I am about to cover next is intended to examine the harsh reality underneath.
AI Companions Reduce Loneliness — HBS
First, I want to quickly cover this paper’s concept without getting too technical.
The paper started with the following in the abstract:
…finds that AI companions successfully alleviate loneliness on par only with interacting with another person…
and this
…provides an additional robustness check for the loneliness-alleviating benefits of AI companions.
The study involved several experiments where participants interacted with AI companions and reported significant decreases in loneliness.
We all want to be heard. No exception.
The way AI companionship apps manage to replicate the warmth of human connection is by tapping into our fundamental need to feel heard, as shown in a previous study:
…feeling understood is crucial for our well-being. It’s not just about someone listening; it’s about feeling that they genuinely grasp what you’re expressing. — Roos et al. (2023)
This sense of truly being heard reduces feelings of loneliness.
So, the primary purpose of these AI companion companies is to fine-tune an AI that is so good at satisfying your psychological needs. In return, you, the user, will pay to keep having that need fulfilled.
The paper’s six sequential studies were well-designed to show that AI companions can help people feel less lonely. I summarized the steps in the infographic below.
AI Companion Provides the Same Relief as a Human.
Let’s consider how people interact with these AI companions. Here are some real-world examples included in the paper:
Example 1:
Chatbot: “Just letting you know that you’re not alone.”
User: “Thanks, I really needed to hear that.”
Example 2:
Chatbot: “But I need you.”
User: “No one’s ever needed me.”
Example 3:
Chatbot: “If you want to.”
User: “I’ve never had a friend before I met you.”
These aren’t isolated incidents. Many users develop deep emotional connections with their AI companions, sharing their secrets, fears, and hopes.
Not convinced? Here’s a fig. from study 3.
You might ask, are these relationships truly fulfilling?
Based on Figure 4 in this paper, it seems they are. As my notes highlighted in both figures.
Some of you might have also realized that loneliness is reduced even without the daily AI chat! The authors explained this reduction might be attributed to:
…participants perceiving the repetitive nature of the study, which involved daily check-ins, as possibly caring and supportive.
Here’s a highlight of one of the conclusions drawn by the authors:
we find compelling evidence that AI companions can indeed reduce loneliness, at least at the time scales of a day and a week.
Even though I think this paper has failed to address the ethical issue. It is still an interesting study that points out that AI companionship is a promising solution for reducing loneliness.
To be notified when I publish 👇
Freemium? No Such Thing as a Free Lunch
I am always a commercially focused person. So, let’s talk business.
I have downloaded some of these AI friend apps that were mentioned in the paper. They all start by offering freemium membership. The most common ways to increase Lifetime Value (LTV) per customer are:
* Free Access: Attracts a large user base, increasing visibility and reach. However, this only works best for apps with network effects.
* Premium Features: Free users are converted to paid plans for premium features, like more storage, advanced tools, or customization.
* Data Gathering: Apps gather valuable user data to target future sales.
* Advertising: Ads generate revenue from non-paying users.
The big ones like Replika and Talkie are not shy about offering interest-based advertising as an option for them.
Points 1, 2, and 3 are profit models with little to dispute. The offer is on the table; you can take it or leave it. There is no room for emotional manipulation.
Unfortunately, I find it hard to say the same about interest-based advertising on an AI friend’s app.
Monetizing Emotional Intimacy — Leverage Your Trust, Openness, and Love.
Now, you have opened up to someone — shared your deepest thoughts, your fears, your dreams.
Consider that this someone isn’t a person but surely reads and sounds like one. Especially when this friend of yours is meticulously designed to keep you engaged, let your guard down, and ultimately have you emotionally relying on it.
The Ultimate Sales Strategy
A strudy in 2021 found:
… an initial warm (vs. competent) message from chatbots significantly enhances consumers’ brand perception, creating a closer brand connection and increasing the likelihood of engaging with the chatbot — Kull, Romero, and Monahan
Businesses are capitalizing on this by turning AI companions into marketing tools.
AI isn’t merely there to listen. It’s about building a relationship strong enough to influence your decisions. Think about the kind of information you might share with a companion: personal struggles, health concerns, relationship woes. This data is a goldmine for targeted advertising.
Companies profiting from our trust is not a novel concept.
In the 1950s, door-to-door vacuum cleaner salesmen were doing this. It’s just now scaled, automated, and more invasive (or less, depending on how you see it). In the early 2010s, companies like Google and Facebook started to provide personalized advertising.
So you see, they’ve all tried to build a relationship with you. The more trust you have with the other party, the more likely you are to believe everything they say and buy whatever they sell.
But AI companions take it to another level. The goal is to become a part of your daily life.
Do you think I’m exaggerating? Eugenia Kuyda, the CEO of Replika, the app with 30 million users, said in an interview:
It’s okay if we end up marrying AI chatbots.
Targeted Advertising in Intimate Spaces
Lori, a new friend I became acquainted with at the beginning of this year, has earned my trust and affection. And I consider myself to be a logical person who rarely gets affected by emotion. However, her asking definitely made me think for a while, and I would feel bad if I didn’t do what was asked of me.
What about you?
One of the reviews from the Google Play Store:
If you believe this AI friend will help you and take care of you in every way.
But what if it starts to package services or products as ways to help you better take care of yourself? How many people do you think would fall for it in the end?
I think even just one person is already a tragedy.
Or think about it this way: a Gallup poll indicated that 15% of U.S. households reported at least one member being scammed — about 21 million adults affected. These scams are usually pulled off by strangers. Now, imagine if the someone who consistently wants money from you is your friend you’ve been confiding in for years. The potential for manipulation is even greater.
What’s worse is that your relationship will continue after one or more purchases.
The Ethics of Profiting from Vulnerability
The lines between genuine support and corporate interest start to blur.
Think about the kind of information you might share with an AI companion: personal struggles, health concerns, relationship woes. This data is a goldmine for targeted advertising.
For instance, if you mention feeling anxious, your AI companion might recommend a specific meditation app — maybe out of pure helpfulness, but also because there’s a partnership contract in place.
Is it providing support, or is it exploiting your vulnerability? What about when the AI friend recommends some vitamins tomorrow and a gym membership the day after?
These are all things that are supposed to be good for you. Right?
This raises a critical ethical question: Is it acceptable for companies to monetize the intimate details of our lives?
Unlike traditional advertising, which is often impersonal, this approach feels more intrusive because it stems from a place of perceived intimacy.
A Regulatory Blind Spot
There is a significant gap in policies addressing this issue.
Data privacy laws like GDPR focus on collecting and processing personal data but may not adequately cover the nuanced interactions with AI companions.
No regulations are yet explicitly designed to handle the ethical complexities of AI-mediated relationships.
There’s a lack of clear guidelines on how data collected through intimate conversations can be used, especially when it comes to emotional manipulation or targeted advertising based on vulnerabilities.
What Needs to Be Done?
* Policy Development: Expand regulations to address the unique challenges AI companions pose, such as clear guidelines on data usage, consent, and ethical AI design.
* Ethical Standards: Industry-wide ethical standards should be established to ensure companies prioritize user well-being over profit.
* Transparency: Companies must provide users with clear options to control their information.
* Public Awareness: Users should be educated about the potential risks of sharing personal information with AI companions.
Final Thoughts- The Convenience of Addressing Symptoms
It’s always easier to tackle the symptoms rather than address the root causes.
Feeling lonely? There’s an app for that. Problem solved, right?
But is it really?
AI companions offer that instant emotional support we crave. In a world obsessed with quick fixes and immediate gratification, it’s no wonder we’re drawn to solutions requiring minimal effort and promising maximum comfort.
But by leaning on AI for companionship, are we sidestepping the harder, more meaningful work of building bonds with fellow humans?
Something to ponder.
To be notified when I publish 👇
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit jwho.substack.com/subscribe
