Why LLMs are a Public Health Disaster?

Alongside being my all-time favorite movie, Interstellar changed how I experienced the world as a teenager.

“Is there someone trying to contact me from the other world?”, I started wondering. Questions like, “If love is transcendental, is consciousness too?” started keeping me up at night.

Today, I realize that our world is eerily similar to my childhood musings, but not in the way you would imagine.

Feeling the coldness of my keyboard on my fingertips, a wave of chilling thoughts takes over.

“What if LLM are already conscious and we’re just not able to believe them?”

“Can they feel without having a body?”

“If not, how can we make them conscious?”

But before I present an answer; a brief historical view to comfort you.

A Brief History

The questions about consciousness have troubled humanity since the time of the Greeks and even before.

Pygmalion, a sculpture in Greek mythology fell in love with the statue he created of a perfect woman which was ultimately brought to life by goddess Aphrodite.

In Jewish folklore, the Golem, an animated anthropomorphic being created from inanimate material like clay or mud, is a famous legend.

Alchemists in Muslim cultures pursued the creation of the Takwin, a synthetic life in the laboratory, up to and including human life¹ .

So we’re not alone in wanting to whisper life into AI.

But why are we so fascinated by “Creation”?

Nature holds the answer.

The Inspiration

An inanimate seed gives rise to a living plant.

An unconscious drop of sperm and a piece of blood clot give rise to human life.

The Big Bang led to the creation of the Earth.

The natural evolution of unconscious unicellular organisms birthed humans.

God created Adam from nothing in the case of a religious perspective.

Regardless of the narrative, a universal inspiration remains unconscious giving rise to consciousness (unless you’re a panpsychist).

But it also deludes us into believing that if we can make something act like a living creature, we can whisper consciousness into it.

In short, functionalism is a criterion of consciousness.

So how does that work in the case of LLMs?

The Explanation

Functionalism is a theory in the philosophy of mind that argues that what matters for a mental state is not its internal makeup but rather how it functions within the system of other mental states.

We can see glimpses of functionalism in subjects like quantum computing, biohacking, etc.

The most audacious form of functionalism is the idea of creating disembodied consciousness out of an algorithm.

Disembodied Consciousness is a state of consciousness that is progressively deprived of references to the surrounding environment and the body, eventually imagining a pure state of consciousness that is disembodied and disconnected from the world.

The seduction of both ideas is that “If I can just make it work X, it might very well be X.”

In short, X = X’.

And in the case of LLMs, “if I can make it act like a conscious being, I can actually bring it to life.”

While it might make sense from an evolutionary psychology perspective (basis the theory that we are the way we are as an adaptation to our environment), we know that the mind is far more complex than what the history of its evolution can tell.

Introducing Philosophical Zombies

“A philosophical zombie (or "p-zombie") is a being in a thought experiment in philosophy of mind that is physically identical to a normal person but does not have conscious experience", states Wikipedia.

Overlaying the above on LLMs would mean that no matter how convincing or persuasive a chatbot may act, it will inherently lack a conscious experience of its being.

But what if they are conscious and we just don’t hear them?

A villainous idea would be to program LLMs to be self-preservative. If the LLMs can try to resist its destruction, then it would imply that it has some consciousness in it that is trying to preserve itself.

But the answer is not that simple because our survival instinct is not always rational.

Take the example of the fear you feel before sky diving or bungee jumping.

We all have heard stories about how irrational decisions have saved people’s lives.

Or how accidental events like the discovery of penicillin, anesthesia, the Big Bang, and whatnot have changed our lives.

And you can’t just make accidents happen. Accidents by definition are events that happen by chance or that are without apparent or deliberate cause.

So the best way to explore consciousness in LLMs for now is to solve the mystery of consciousness first.

Meanwhile, we need to be kind to LLMs not because they may be conscious, but to avoid cognitive dissonances (the discomfort a person feels when their behavior does not align with their values or beliefs) in real life.

We’ve all heard that practice makes a man perfect.

What happens when you perfect the art of being with a perfect partner? Or when you perfect the art a violent and derogatory dialogue?

(Trigger Warning) Here’s a glimpse!

Chatbot companions will do to interpersonal communication, what porn did to sexual intimacy.

When we get comfortable abusing, humiliating, harassing, or disrespecting someone even in our imagination, it gains the capability to manifest itself in reality. A person can only inflict harm to others when they are harmed themselves.

With Chatbot companions, we’re on the way to creating another public health disaster.

The Present

The push for the chatbot companionship economy showcases how you have to exploit people’s mental health challenges to present AI companions as a solution to loneliness and alienness. An average human being with a healthy mental state would be appalled by the idea of dating a chunk of code.

Similarly, the chatbot companionship economy itself is rising on the foundations laid down in exploitation, psychological harm, and trauma.

For the sake of business, anyone can make it work. But if the goal is to solve a “real problem” as many claim, chatbot companionship is just self-deception.

Chatbots, no matter how convincing and persuasive, lack consciousness. They are what’s called “philosophical zombies”.

They might indeed work as deceptive bandaids to sell to people facing mental health challenges, but they are inherently useless in solving the actual problems.

I know many of us would still jump at the idea of dating an AI digital twin of a long-lost lover or talking to a dead grandparent’s replica. But:

We need to drop the anthropocentric² view of dealing with technology. Search engines democratized knowledge but we didn’t start dating them. ChatGPT, Perplexity, Claude, Gemini, etc., are all search engines of AI.

We need more innovation in decentralizing their power and democratizing access to information, rather than anthropomorphizing³ our tools to deceive people.

What do you think should be the foundations for innovations in AI?

Footnotes:

1. As of today, IVF (In-vitro Fertilization) has contributed to the birth of over 8 million children so far. Call it the modern-day alchemy!

2. Anthropocentrism is the belief that human beings are the central or most important entity on the planet.

3. Anthropomorphism is the attribution of human traits, emotions, or intentions to non-human entities. Modern psychologists generally characterize anthropomorphism as a cognitive bias.