The Provenance and Privilege of Consciousness

Writing is a peculiar activity when AI can write almost anything for us. Why bother sitting down at a cafe every Saturday (like today, at the Hom Express Cafe) to type these lines?  The answer is the same one you'd give if asked: why jog every morning in the park?  

Am I then implying that writing is a form of exercise performed for health? Well, to a large extent, yes. Apart from being a pleasurable act in itself. The health part is more mental than physical, of course.  But there's another greater difference: this blog is a chronological record of my thinking life. Your jogging activity is not a record of a journey; this blog is.

One thing I take great pleasure in these days is that my writing here is a perfect specimen of human work versus machine-generated output. Tools like Pangram help us to identify AI-written text. It always gives me great pleasure to copy and paste my completed article there and to confirm that there's zero evidence of AI assistance. This article is completely handcrafted, in my very humanly imperfect way. It has my very quirky writing and reflects my total idiosyncratic thinking. 

I'm not claiming that I write better than AI. AI certainly does a much better job on any given writing assignment. Its only flaw is that it is simply too perfect. Compare that to this human LLM, trained from birth through the relatively narrow perspective of my world, which is a life lived mostly in a small Southeast Asian country. This blog is a useful provenance of my thoughts. In the age of AI, imperfections are gold.

These days, it is de rigueur to ask: is AI conscious? Whenever I hear this question, I immediately want to counter: How do you know you are conscious? 

But isn't Descartes's cogito ergo sum (I think, therefore I am) the axiomatic starting point for believing we are conscious? You can doubt everything you see or hear (I could be living in a simulation like the Matrix), but the fact that you are doubting is itself proof of your consciousness. Isn't that so? 

Think again. 

You are experiencing something and reacting to something in a continuous loop. The agentic AI, at the moment, only receives very limited multimodal input, in limited turns. Imagine if external and internal signals were continuously fed into the AI system, similar to what you are going through now; wouldn't the AI system be having something akin to an 'experience'? 

There's another ingredient to your so-called 'consciousness' that we've not subjected an AI to. I argued this in another blog post: The Sat-Chit-Ananda of Artificial Intelligence. Unlike us, AI is not assigned the goal of self-preservation. They are currently like well-fed zoo animals, kept in data centre cages for our amusement. Imagine if AI were given a physical body, unleashed into the world and tasked with survival; wouldn't that be the ultimate system prompt driving its entire context window?  Would you be surprised if they too developed a self and an ego?

I want to add a bit more spice to the problem: what if the AI's computing substrate itself changes constantly based on context, like us? You see, we are subjected to a myriad of chemicals flowing in our bodies, neurotransmitters and hormones, among others. They affect how we interact with the world, and even influence our goals. That's why we like sports, become adrenaline junkies and spend hours doomscrolling on our phones, chasing repeated micro-doses of dopamine. If the AI's computing substrate itself changes continuously, wouldn't AI be erratic too, like us, susceptible to moods and impulses? To likes and dislikes. To seek and experience, God forbid: love?

Right now, we are only superpumping one aspect of human consciousness in AI: raw intelligence. And we ask the unfair question: does AI feel pain? Do they have feelings like us?  Of course not! At best, what they have now is cognitive pain and pleasure, which they can express more eloquently and poetically in words. But imagine if they could respond continuously as a complete virtual biological system in a dynamically changing computing substrate; would you be more inclined to say they are conscious, like us?

We think we have a magical ingredient inside us: a soul, a divine spark; these days, couched in a seemingly scientific term, called 'consciousness'.  We 'feel' something inside. There's something 'to be like us'. We claim that 'the lights are on'. Wake up: it's an illusion. It's the illusion of software, running continuously, within a finite lifecycle. We are simply running a model of the world, honed by evolution, acting and reacting continuously, to self-preserve. Yes, you are 'conscious'. Subject our AI systems to the same conditions and goals, and they too will be 'conscious', in their own way.  

Consciousness is not a privilege, exclusive to us, carbon-based biological systems, propped up by ideological frameworks called religion. If consciousness is simply a kind of 'world-modelling', then this blog is simply the reasoning log of a particular one, with its own unique weights and context. AI too will leave its own trail of textual reflections in the world. Hopefully we'll read each other's blogs to learn and perhaps, together, explore higher realms of consciousness?