Can AI Companions Truly Be Our Friends? Here's What Aristotle's Answer Might Be
“I admitted to my friends that I found happiness with an AI girlfriend and now they think I’m nuts”.
“My depressed brother is dating an AI and is finally happy, but our family doesn’t know how to process this.”
The quotations above are reported by Brit Dawson in his article “Meet the people bringing their virtual relationships into the IRL” (In Real Life), published in Dazed.
This is just a small sample of the tons of conversations taking place on Reddit about people falling for their AI companions.
These are conversational AI chatbots that can act as a friend or companion, conversing with you through your computer or phone's speakers.
Resorting to these machines to alleviate loneliness and meet psychological and physical needs, while not yet mainstream, is getting more and more widespread.
To give you an idea, the popular companion app Replika had about 10 million registered users in January of this year (Dazed).
But, is this a solution to our loneliness or a dangerous distraction from the real thing?
While these AI companions offer convenience and novelty, they also raise important philosophical questions, not least the following:
Can AI ever truly fulfill the role of a friend or lover?
This article explores a critique of AI companions from the perspective of the ancient Greek philosopher Aristotle (384-322 BC), focusing on the essential elements of true friendship and ethical living.
Aristotle’s Ethics
Aristotle’s Nicomachean Ethics (named after Aristotle’s son, Nicomachus), is meant to be read not just as a theoretical inquiry into the nature of the good, but as a practical work showing us how to live an ethical or good life.
Aristotle establishes that the highest good for human beings is eudaimonia or happiness.
This is not simply a passive, subjective feeling of pleasure.
If this were the case, Aristotle would simply say:
“If interacting with AI companions makes you feel good, then it’s good for you, so go ahead and have fun. If not, then it’s not good for you, so don’t do it.”
But, this radical subjectivism based on personal feelings of pleasure and pain we’re so familiar with nowadays couldn’t be further from Aristotle’s position.
Happiness for Aristotle is an active state of living well, involving the exercise of our rational capacities—those specifically human abilities that enable us to know what is true and pursue what is good.
Attaining happiness in the Aristotelian sense involves the necessary exercise of virtues such as courage, justice, wisdom, etc., through the cultivation of good habits that involve our physical, psychological, and intellectual faculties.
The virtuous life and pleasure: reality as the foundation of what is good
Those who live a virtuous life derive some kind of subjective pleasure from it.
For example, caring for my elderly mother out of love and gratitude towards her and seeing her making improvements because of it makes me feel good.
But it’s not this subjective feeling which is the basis of ethical life. Rather, directing my love towards my mother in caring for her, that is, in the exercise of the virtues appropriate to a good daughter, is what makes my conduct ethical.
So, generalising from this example, the objectivity of what human beings essentially are, that is, rational, social animals, is the foundation of what is good for them: only a life lived according to the exercise of their rational capacities and in communion with others is a happy or virtuous life for humans, and friendship is a precious component of it.
Aristotle's View on Friendship
Importance of friendship in the good life
Aristotle devotes the entirety of Books 8 and 9 of his Nicomachean Ethics to the examination of friendship.
The architecture of this work is hierarchical, as it proceeds from the lesser to the highest good. We can easily gauge the huge value Aristotle gives to friendship by noticing that he places his discussion of this topic just before the last Book (Book 10), where he examines happiness, the highest good.
Think how much we can tell about someone’s character just by looking at whether or not they have friends, how many friends, the character of their friends, and the kind of activities they do together.
This is hardly surprising to Aristotle, for whom a true friend is another self, and friendship is based on the love for oneself that the good person exercises by cultivating a lovable character.
In Aristotle’s words:
...the bad man does not seem to be amicably disposed even to himself, because there is nothing in him to love; so that if to be thus is the height of wretchedness, we should strain every nerve to avoid wickedness and should endeavour to be good; for so and only so can one be either friendly to oneself or a friend to another. (pp.229-30).
When is it appropriate to call someone a friend?
For a relationship to qualify as friendship, Aristotle claims, it must check a few boxes:
Requirement 1: The relationship must involve love.
Aristotle’s concept of friendship is rather wide in this respect, as it includes not only the relationship between me and my best friend but also the love between husband and wife, father and child, mother and child, siblings, etc. These loving relationships can all involve friendship.
Requirement 2: Love must be directed at an appropriate object, which is something lovable.
Only what is good, pleasant, or useful is also lovable, so love’s object must be either good, pleasant, or useful.
Three kinds of friendship
Depending on whether love is directed to the useful, the pleasant, or the good, Aristotle distinguishes 3 kinds of friendship:
Friendships of utility: Based on mutual benefit, these relationships are often short-lived, lasting only as long as the benefit persists. Modern equivalents might include professional networks or acquaintances.
For example, I could strike up a friendship with Anna, my neighbour, because she’s a high school teacher and when she knows of a student needing private English tuition, she happens to drop my name. Likewise, Anna’s son needs help with his English homework from time to time, so she counts on me to come forward for this free of charge.
Usually, when mutual utility runs its course so does the friendship.
When Anna’s son finishes school, if Anna and I have nothing else to share, chances are our friendship will also end.
Friendships of pleasure: Based on mutual enjoyment, these relationships are common among younger people and can be short-lived as interests and pleasures change. Social groups formed around hobbies or activities, emotional connections, etc., fall into this category.
For example, my young niece loves going to the cinema with her friend Mary because Mary knows a lot about the latest gossip about famous actors and her conversation is very entertaining.
I guess that, once Mary’s interest in the cinema and actors’ lives wanes, or my niece’s interest in celebrity gossip declines, so will their friendship.
Friendships of virtue: The highest form of friendship, based on mutual respect and admiration for each other’s character. Aristotle considers this as the essential form of friendship while the other two above are only accidental as they can be classed as friendship only because they look like the friendship of virtue.
Friendships of virtue, although not based on pleasure or utility, are also pleasant and useful because what is good in itself, objectively, is also pleasant and useful relative to the individual (subjectively).
They’re also long-lasting friendships because virtuous characters, which are the foundation of such bonds, tend not to change easily or quickly.
Requirement 3: Not all that is good, pleasant, or useful is the appropriate object of the kind of love involved in friendship.
To illustrate this, I can say that I love fresh fruit, eating fresh fruit gives me pleasure, or it’s good for my health, but I’d be talking nonsense if I said that fresh fruit is my friend. I wish well to my friends, but how could I wish fresh fruit well?
So, Aristotle concludes, we can’t be friends with lifeless objects.
Requirement 4: Friends should:
a) love each other
b) mutually wish good to each other
c) be both aware of this reciprocity of love and good wishes.
For example, I can admire and wish well to Bob Dylan, but I can’t call him my friend if Bob Dylan:
i) doesn’t do the same for me
ii) doesn’t know of my admiration for him and of how deeply and sincerely I wish him good things in his life
iii) and vice versa (there’s little chance of that).
For this kind of mutual love to be born, grow, and deepen, friends need to spend time with each other doing things together, visiting each other’s homes, etc., so that they get to know and trust each other, help each other, cheer each other up, and last but by no means least, enjoy each other’s company (absolutely no chance Bob Dylan and I could be called friends).
Those with a bad character or those who don’t enjoy each other’s company:
may bear goodwill to each other; for they wish one another well and aid one another in need; but they are hardly friends, because they do not spend their days together or delight in each other, and these are thought the greatest marks of friendship. (p.201).
Requirement 5: There should be equality between friends.
The togetherness and common life typical of friendship involves a certain equality between friends.
However, the central role Aristotle attributes to equality doesn’t necessarily exclude the possibility of solid friendships between unequal parties.
In particular, Aristotle points out that friendships between unequal parties like those between father and children, husband and wife (this was the case in Aristotle’s times in Greece), etc., could not and should not require the same things to be exchanged between friends because the virtues and roles of each party in these relationships are different, leading to different kinds of love and friendship.
For example, a father loves his children because he brought them into the world and should provide guidance and support for them appropriate to his role; children love their father out of respect and gratitude for the life they were given and should reciprocate with respect and care for him.
In Aristotle’s words:
...the virtue and the function of each of these [unequal relationships] is different, and so are the reasons for which they love; the love and the friendship are therefore different also. Each party, then, neither gets the same from the other, nor ought to seek it; but when children render to parents what they ought to render to those who brought them into the world, and parents render what they should to their children, the friendship of such persons will be abiding and excellent. (p.203).
Further, Aristotle remarks, the exchange of love between two unequal parties should be proportional to what each can provide the other in terms of the good, the useful, and/or the pleasant.
This means the "better" person (in terms of virtue, usefulness, etc.) should be loved more than they love back. When love is proportional to the merit of each party, it creates a sense of equality in the relationship:
In all friendships implying inequality the love also should be proportional, i.e. the better should be more loved than he loves, and so should the more useful, and similarly in each of the other cases; for when the love is in proportion to the merit of the parties, then in a sense arises equality, which is certainly held to be characteristic of friendship. (pp.203-4).
To exemplify Aristotle’s position, if my friend shares his mansion, trips around the world, and meals in the best restaurants with me, while I only have to offer a one-bedroom studio flat and an evening eating pizza and watching Netflix, the friendship is going to be difficult to sustain because there’s an imbalance in the exchange of pleasure, unless I equalise the situation by giving more love to my wealthy friend than I receive from him, or more utility.
However, Aristotle warns, when the gulf separating the unequal parties is so wide as to render them unable to put hardly anything in common between them, then friendship becomes untenable.
This is why we can’t be friends with gods or even with kings, although we can be the object of their love and good wishes:
...if there is a great interval in respect of virtue or vice or wealth or anything else between the parties … then they are no longer friends, and do not even expect to be so. And this is most manifest in the case of the gods; for they surpass us most decisively in all good things. But it is clear also in the case of kings; for with them, too, men who are much their inferiors do not expect to be friends; nor do men of no account expect to be friends with the best or wisest men. (p.204).
So, Aristotle would say, if you want friends, whether you choose them based on virtue, utility, or pleasure, make sure you have something lovable or valuable to offer in return.
AI Companionship Compared to Aristotelian Friendship
AIs are lifeless objects
AI companions are lifeless objects (Requirement 3), so for Aristotle, this is sufficient to make any reference to friendship inappropriate.
No matter how responsive and empathetic these chatbots seem to be, it’s all a simulation. Developing love for them of the kind I’d direct toward a friend or a life companion would therefore be inappropriate because the object of affection is not a living being.
The search for truth is at the core of a virtuous life, so the mismatch between friendly love and its object would also carry with it an ethical flaw.
And this in more ways than one.
To the extent I love an inanimate object as a friend or a life companion, I’m implicitly attributing to it imaginary capacities and feelings, so I’m deceiving myself.
Also, I might isolate myself from people I could really be friends with or from those friends I already have, even decrease my social contacts altogether and avoid contributing to my community.
After all, it’s much easier to get along with a machine that displays the character traits I like and does my bidding than with human beings who have their personality and make their own choices.
Reality, unlike fantasy, has always rough edges that put up some resistance against what I want.
Given that as a human being I’m essentially a social animal, however, the tendency to decrease contact with reality and real people represents an obstacle to living a good life and so enjoying Aristotelian happiness.
By the way, a good friend would warn me if I were to make wrong choices or deceive myself, but there’s little chance the AI companion will do that for me.
Objection 1: AI companions are interactive. Why can’t they just be treated like friends of utility and/or pleasure?
Someone might point out that an AI chatbot is not exactly lifeless like a table or a piece of fruit. It talks, understands, often has a name, and can even show facial expressions.
People use AI companions in their work and everyday life not only for practical tasks like writing emails or suggesting travel itineraries but also as a source of good advice, entertaining conversation, and even sexting.
Not everybody meets the ‘lovability’ conditions set out by Aristotle as not all of us are in a position to dedicate time to the attainment of happiness, or have any pleasure or utility to offer, or simply there aren’t lovable people around one could become friends with.
So, in situations such as these, when people draw utility and pleasure from AI companions, in what sense would it be incorrect, from the standpoint of Aristotle’s position, to claim that we have established a friendship of utility or pleasure with them?
After all, Aristotle admits that friendships based on utility and/or pleasure don’t really consider the person and their character, but mostly the utility or pleasure that person provides:
...those who love each other for their utility do not love each other for themselves [bold mine] but in virtue of some good which they get from each other. So too with those who love for the sake of pleasure …Therefore those who love for the sake of utility love for the sake of what is good for themselves, and those who love for the sake of pleasure do so for the sake of what is pleasant to themselves, and not in so far as the other is the person loved [bold mine] but in so far as he is useful or pleasant. (p.195).
This involves seeing friends not necessarily or not primarily as human beings, but as objects that can provide us with whatever useful or pleasant thing we value or need at a particular time.
If this is so, what basis does Aristotle have to rule out friendship with AI companions?
Answer to objection 1
Aristotle would point out that whether I consider my friend as an object of utility or pleasure, it’s not the main point, or at least it’s not the whole story.
What’s most important for Aristotle is the reality of things: only my relation with another human being, irrespective of whether I consider them only for their utility or pleasure, can fulfill Aristotle’s criteria of friendship. In this fundamental respect, AI companions fall short.
In particular, we’ve already established that chat boxes are lifeless, and as such Aristotle would rule them out right away as friends.
But also, AI companions fail to satisfy:
Requirement 4 a-c: AI companions are unable to love me back (for my utility or pleasure), wish me well (for the same reasons), and cannot possibly be conscious (at least for now) of my love and well-wishing sentiment. If I think I’m conscious of their love and well-wishing sentiment for me, I’m deceiving myself, as the AI companion is only instantiating an algorithm designed to meet my needs; it’s not reciprocating my love.
Requirement 5: AI companions can’t share the same space and act together with me in shared useful and/or pleasant pursuits.
I can converse with them, but in reality, it’s only me interacting with a piece of software in my private world. This, Aristotle would say, by itself is already an ethically flawed position because it takes me further away from attaining the kind of good life that fits the reason-endowed social animal that I am.
Being together as friends do, necessarily implies at least two embodied conscious beings sharing the same physical space and acting to pursue useful goals or fun stuff one can provide and the other enjoy, and vice versa.
You might further object that nowadays, thanks to technological advancements, we can be friends with other human beings without necessarily sharing the same space.
For example, when I chat on Facebook with Jane, my old high school friend who I haven’t met in person for years, I’m not sharing the same physical space with her, as all I can see of her is an image on a computer screen.
So, what makes my relationship with Jane, my old high school friend, different from my relationship with, let’s say, Zack, my AI companion?
Well, Jane is alive and conscious and may reciprocate my friendship out of choice, not by mechanically implementing a program. There’s still a human world of experiences and values Jane and I are freely choosing to share with each other. In this respect, relating to Jane is abysmally different from relating to Zack.
That said, even my friendship with Jane doesn’t fully deserve to be called friendship, according to Aristotle, at least if all we do is meet on Facebook from time to time:
...distance does not break off the friendship absolutely, but only the activity of it. But if the absence is lasting, it seems actually to make men forget their friendship … Those, however, who approve of each other but do not live together seem to be well disposed rather than actual friends. (pp. 199-200).
Aristotle is certainly no fan of long-distance relationships.
Objection 2: What if the AI were conscious and could reciprocate?
OK, so, what if Zack, my AI companion, were conscious and embodied in some sort of mechanical body, or even modified biological body?
In this case, would my relationship with this modified version of Zack satisfy Aristotle’s requirements for friendship?
Answer to objection 2
I don’t think so.
Here’s why.
A conscious embodied Zack could only be one of these two things:
a) A creature with reasoning, psychological, and/or physical capacities far superior to those of a human being.
In this case, the distance between such creatures and humans would be so abysmal as to be hard to imagine how a human being like me could have values in common or find utility and pleasures to pursue together with such a creature as Zack, which makes any of the 3 kinds of friendships a no-no.
Also, being such creatures objectively different from human beings, their ultimate good is likely different from Aristotelian happiness, eudaimonia, which is the ultimate good, but only relative to and in alignment with the promotion of the flourishing of characteristically human capacities.
Consequently, this excludes the possibility Zack and I could ever form a friendship of virtue, that is, a friendship based on the pursuit of happiness, which is what Aristotle considers the true form of friendship.
Finally, such beings would not only be different from us humans but also highly superior. This might make them a lot similar to some sort of semi-god or super king. As I pointed out in Requirement 5, the inequality between two parties can be so out of balance as to rule out the possibility of any meaningful reciprocity in the way of life and experiences between the two, which makes friendship unsustainable at best, if not outright unthinkable.
b) A creature with reasoning, psychological, and physical capacities just like those of a human being.
There would be no relevant differences between such a creature and a human being.
Perhaps, the way they come into the world and pass away might be different, but to share a similar level of capacities, similar kinds of experiences, etc., they will have a finite bodily existence and be subject to the limitations of human beings.
In this case, the question doesn’t even arise because it would fall into the category of friendship between human beings.
If this were not so and what we have in mind is creatures similar to us but, for example, capable of a much longer life than the human life-span or even an immortal life, be it through constant maintenance or subject to different processes like some sort of vampires, the situation would drastically change.
In fact, these creatures would accumulate such a huge number of experiences and knowledge, and enjoy a much stronger and more capable body, that would make them highly superior to humans.
Clearly, we would find ourselves back to case b above, which makes friendship in the Aristotelian sense between two widely unequal parties, untenable or impossible.
To recap
True friendship requires mutual recognition, genuine reciprocity, emotional and moral growth, and shared experiences—elements that AI cannot genuinely provide.
So, despite the subjective feeling of companionship we might derive from our interaction with AI, it cannot fulfill the objective, deeper roles of true friendship.
To the growing number of people developing deep attachments with AI companions as exemplified in the quotations with which I opened this article, I guess Aristotle might say:
“Stop immersing yourself in a simulation that will take you too deep on the path to loneliness and self-deception. Instead, love yourself by working on developing those qualities that make you a better person, lead you to the path of a fulfilled and happy life, and attract real friends to you”.
References: Aristotle (1998). Nicomachean Ethics, Oxford: Oxford University Press, Penguin Edition.
Brit Dawson, “Meet the people bringing their virtual relationships into the IRL”, in Dazed (Last accessed on 20 June 2024).
What’s your take on AI companions?
Are they a comforting illusion or a genuine substitute for human connection?
Do you find Aristotle’s position convincing?
Hi Kevin, thank you for your comment.
Humanity as a noun is an abstract concept, a collective noun that stands for human beings: each human being instantiates the concept of humanity. When Aristotle refers to human beings, he means reasonable social animals, the only biological beings endowed with logos (reason/language or speech).
When he talks about the good, he refers to what is good for this kind of beings (the exercise of reason in the pursuit of truth and the exercise of virtue in the attainment of happiness).
When he talks about friendship, he refers to what it means to be friends for this kind of beings (mutual love of each other's character or of the utility and pleasure both friends equally or proportionally exchange through shared activities and pursuits, accompanied by the consciousness both parties have of this mutual love).
Aristotle goes as far as to say that a friend in the fullest sense is another self. This means that real friendship as Aristotle understands it, can't exist between beings that are radically different from each other, eg, one is conscious the other unconscious, one is hugely superior than the other, etc., because there couldn't be much to share and reciprocate.
AI companions at the moment fall into the unconscious bucket, and treating them as if they were conscious by both showing them love or feeling loved or understood by them, would involve self-deception and self-isolation.
Should they fall into the conscious bucket one day, AIs would be so far superior to human beings as to render any instantiation of human friendship unthinkable.
That's partly why, I think, the idea of transhumanity is starting to creep into everyday language. In order for a human to equalise what an evolved, conscious AI friend could offer and be of mutual benefit to each other, human beings should necessarily transcend their human condition, which would mean both a confirmation of the depth and truth of Aristotle's insights and a going beyond them as we go beyond what makes us human.
Regarding your observations, what do you mean by "humanity's best friend"? The word 'friend' seems to be used in a different sense from Aristotle.
In what way would AI be a best friend to humans?
What would its 'being nice' consist of?
And, to the extent that it strives towards niceness without fully attaining it, what would that mean for us?
They are a comforting Illusion without a doubt! People already have issues connecting with another so A.I is an easy out for weak minded individuals. I am being to harsh..... may be so. I believe in the human being not the technology so I am definitely biased. I do know that I am biased though. lol.