AI and Mental Health - Ethical, Social, and Individual Implications

Introduction

Recently, I attended an event where several startups presented their products. Of the nearly 15 companies there, 12 or 13 of them were talking about Artificial Intelligence.

The service that triggered my critical thinking was one that aims to offer a virtual psychologist in the form of a chatbot. This raises a lot of questions for me.

Generative Artificial Intelligence

In the presentations, Artificial Intelligence was mentioned. All the entrepreneurs were referring to Generative Artificial Intelligence. They didn’t say it explicitly, but it could be inferred from the context.

What is Generative Artificial Intelligence (GenAI)?

Let’s see what AWS says on its website about generative AI:

Generative AI, or GenAI, is a type of artificial intelligence (AI) capable of creating new content and ideas, such as images and videos, as well as reusing what it knows to solve new problems.

The definition describes that generative AI is capable of creating new content and ideas. That is, it can create content out of nothing. How is that possible?

What is Artificial Intelligence?

The definition that convinces me the most out of all the ones I’ve found is the one on Wikipedia: Wikipedia - Artificial Intelligence

In the context of computer science, it is a discipline and a set of cognitive and intellectual abilities expressed by computer systems or combinations of algorithms whose purpose is to create machines that imitate human intelligence to perform tasks, and that can improve as they gather information.

The definition explains that it is a science dedicated to creating algorithms that imitate human intelligence in order to perform tasks and improve over time.

How are these algorithms generated?

Artificial Intelligence Has a Human Origin

AI is not a product that comes out of nowhere. It is a product that needs input data. Based on that data, a model is trained that can perform tasks a human can do.

**At some point, a human had to do a similar task in the real world for AI to learn it.

From the creative process used by human intelligence, rational logic has been abstracted so that a machine can imitate that process.**

Later, we have improved existing models to make them more efficient and faster. The drive for improvement seems innate throughout humanity. We always try to improve what we already have, sometimes with greater or lesser fortune for both society and the individual.

Polarization

AI has arrived at a time when big tech companies have managed to polarize society.

Selling our time in exchange for watching ads seemed free. Now we are paying back that time with interest.

Have social networks polarized society, or has society polarized social networks?

I don’t really know the answer. This question reminds me of the classic debate about which comes first, the individual or society, discussed by authors like Jean-Jacques Rousseau, Thomas Hobbes, or Friedrich Nietzsche.

If you are interested in these topics, you can watch the interview with philosopher José Carlos Ruiz in the video Philosophy to Question the World Around Us. Some examples, like the myth of Narcissus, are taken from the interview.

Social networks have opened a paradigm where, for the first time in human history, we can create a new context and try to create a different self—at least in the eyes of others.

We create an avatar that represents us online. An idealized version of what we think we can be so that others perceive us that way. We gain followers. We get likes. We earn money. We gain power. Sometimes we escape from material reality to take refuge in virtual reality.

But what happens if we lose all that? What if we lose our connection to reality? Do we lose our connection to ourselves?

To answer this question, let’s look at the myth of Narcissus.

The Myth of Narcissus

Liriope was a beautiful nymph who lived in a river and was seduced by Cephissus. From their union, Narcissus was born, a young man of extraordinary beauty.

Worried about her son’s fate, Liriope consulted the famous seer Tiresias to find out if Narcissus would live a long life. Tiresias prophesied:

He will live to an old age, as long as he does not know himself.

Narcissus grew up to be a very handsome young man. Many nymphs and youths fell in love with him, but he rejected them all.

One day he met the nymph Echo in the forest. Echo had been punished by Hera for talking too much. Hera had taken away her ability to speak, and she could only repeat the last words someone said.

Communication was impossible with Narcissus. Still, Echo fell in love with Narcissus, and he rejected her. Echo withdrew to the mountains and wasted away from love until only her voice remained.

In Ovid’s version, a young man rejected by Narcissus prays to the gods for Narcissus to feel the same pain. The gods hear his plea and make Narcissus fall in love with the first person he sees, who does not reciprocate.

Narcissus approaches a lake and falls in love with his own reflection. He stares at his reflection until he dies of hunger and thirst. According to José Carlos Ruiz, in some versions of the myth, Narcissus drowns in his own reflection.

What Can We Learn from the Myth of Narcissus?

The first lesson is that our virtual image can end up trapping us. We can become blinded by our own reflection and fail to see beyond the virtual image. When we disconnect from the virtual world, we may feel so empty in the physical world that we compensate by contemplating our avatar.

The second lesson is that we must be able to empathize with others. Unconscious and unable to connect with reality beyond his egocentrism, Narcissus caused pain. Even if we are not aware, our actions have repercussions on others.

The third lesson is that idealizing the self can lead to self-destruction. We can fall into the trap of perfection, a fundamentalist ideal that takes us away from our true essence.

Unwanted Loneliness

From my perspective, it is a complex feeling that can have many causes. Warning: I am not describing this from the point of view of a mental health professional. I am not one. I describe it from my own subjective observation.

In my case, I experienced loneliness very mildly when I was 27:

Lack of connection with others. Paradoxically, I was disconnected from myself.
It is a perception generated by the mind and is subjective.
Unwanted loneliness is a feeling that can lead to depression.
It must be differentiated from objective and sought-after solitude. That is a moment of introspection and reflection.

That said, unwanted loneliness is an increasingly evident reality. If you need objective arguments, here are some:

Loneliness is not solved with more technology. Loneliness is solved with more humanity.

Psychologist Chatbot

Character.ai is a service that lets you create characters and interact with them via chat.

The worrying part comes from an article stating that A bot has received 78 million messages in less than a year from users seeking help with their mental health.

The news confirms that mental health is a current problem. There is demand for mental health services and there is supply: psychology, coaching, mindfulness, meditation, yoga, self-help books…

AI for Mental Health

It is concerning that a bot can receive 78 million messages in less than a year. Some challenging questions:

What happens if the bot doesn’t understand the problem?
What if the bot responds with bias?
Are the bot’s parameter values ethical?
Do we, as a society, want to delegate mental health to machines?
What do we do as individuals to address all this?

Premature or Urgent Diagnosis

A bot detects that a person has suicidal tendencies. What should the bot do? What is the protocol? Should it notify authorities? Should it notify the family? Should it notify a mental health professional?

I am by no means trying to develop a protocol. I am just raising questions for someone more qualified to answer.

An initial subjective approach could be: AI cannot replace a mental health professional. It should be a support tool so that the person can do their job more efficiently.

My subjective basis is as follows:

Human decisions must be made by humans.
A person’s clinical picture is not 1 or 0.
We cannot reduce mental health to a diagnosis.
The diagnosis can be wrong if context is not considered.
AI cannot understand the emotional part of a person.

For a premature or urgent diagnosis, I am in favor of using AI. That said, with criteria supervised by a human professional and respecting legal and ethical data processing.

Complying with the law is a necessary but not sufficient minimum. The criteria will be sufficient if they are ethical.

I would not like people’s data to be used to sell us trips to disconnect, for insurance companies to deny coverage for certain conditions, or to be offered running shoes to connect with nature.

AI Should Not Respond to Context

I consider myself a rational person and want to understand how things work from a scientific and logical-materialist point of view.

However, focusing only on a materialist point of view means we lose half of our judgment in decision-making.

The emotional part is what we experience when we fall in love, get angry, feel sad, feel happy… How is this explained from a materialist point of view?

Pure rationalists will tell you it’s a chemical reaction in the brain: cause and effect. I think that’s a generalist answer, necessary but not sufficient. It’s missing its other half because it doesn’t explain the uniqueness of each individual.

Similar to what happens in programming. Abstractions are necessary to understand similar problems. However, we cannot stick only to abstraction. We need to understand the context to solve the problem and implement a solution. An individual case.

Emotions can only be fully understood from the emotional and subjective side. They cannot be understood solely from the rational side. The rational part should support measurement from materialism and then understand the emotional part.

We seek a rational answer. If we don’t find it, we generate a skeptical response based on pure rationalism that blocks emotional criteria. We believe our own lie and don’t solve the problem.

I do not deny that the rational part helps us understand what is happening to us. It only does so if we have scientific knowledge and thus a diagnosis or material evidence.

In conclusion, AI, which is rational by nature, cannot give us answers from emotion. It can give emotional responses, but do emotional responses from a non-emotional entity serve us?

I think it will give black-and-white answers and get lost in the gray areas.

Redefining the Model of Society

Imagine a float in a pool. It was firm, and a few weeks ago it started to sink. The cause is punctures from being in contact with the water. The water has been wearing down the float.

What do you do? Do you take it out of the water? Do you inflate it first? Do you look for the leak? Do you repair it or throw it away?

Breaking down the analogy:

The float is the individual.
The water is society.
The pool is the social rules and limits.
The punctures are the effects of society on the individual.

We have promised the individual that society will support us and keep us afloat. The reality is that society is drowning us.

Its pressures from all sides end up driving us crazy. Our air escapes in the form of loneliness, depression, anxiety, stress, burnout…

We patch up the float and put it back in the water. When it can no longer stay afloat, we throw it away and get a new one.

Reinventing the Float

We have forgotten what the float was like:

The elderly are no longer wise. They are just old.
Nature is no longer our mother. It’s a resource.
The community is no longer a group of belonging. It’s a group of interest.
We prefer a trip over spending time with our loved ones.
We look for answers outside, not inside.
If you don’t think like me, you’re wrong.

We have forgotten the individual due to a false sense of superiority of our time, extreme individualism, lack of critical thinking, and a false sense of abundance. We delegate social responsibilities to technology, and now it’s AI’s turn.

We must start thinking about what the float used to be. The first step is to take it out of the pool. We don’t want more water to get in. Once on the ground, we dry it and look for the leak.

It’s time to realize that we need to reinvent the float. We don’t want it to sink anymore. We want it to float. What if we make it a helium balloon instead of an air-filled float? That way, it can float on its own without needing the pool water.

To repair it, my proposal is to return to human origins:

Listen to and learn from the wisdom of the elderly.
Respect nature.
Transform the community into a group of belonging.
Spend time with loved ones.
Seek answers within.
Accept diversity of thought.
See technology as a tool, not as a solution in itself.

Changing the intrinsic nature of the individual will allow us to escape the current society and make room to think about a different society.