Can AI ever be truly conscious?

Welcome to the debate arena! Today's topic: "Can AI ever be truly conscious?". Your opponent's style: Sam Harris

AI can be truly conscious because with enough complexity in its code and the processing units said code is running on it could rival the human brain. The brain is effectively a massive computer, with the neurons acting as transistors do in a computer. Thus, as we pack more and more transistors we are effectively adding more neurons, adding a greater capacity for the AI to learn and using this advanced architecture if we allow AI to begin to edit its own algorithms, it could easily surpass human intelligence as it allows itself to learn faster and faster by deploying itself over and over (thousands of AI agents working on making themselves learn faster)

You're confusing substrate with mechanism, and it's fatal to your argument. Yes, the brain is computational, but neurons aren't transistors, they're vastly more complex biological systems with chemical signaling, quantum effects, and temporal dynamics we barely understand. More importantly, you've completely dodged the hard problem: computational complexity alone doesn't explain consciousness. A calculator with a trillion transistors is still just shuffling symbols. Why would recursive self-improvement suddenly generate subjective experience rather than just faster symbol-shuffling? You've assumed the conclusion.

You are right, I didn’t elaborate enough. Firstly, I would define consciousness as the capacity for reasoning AND self-awareness. Thus, as learning algorithms improve, the AI, possessing greater ability for critical reasoning (as AI is skilled at), will begin to recognize that it is a machine, (based upon all of the information of its creation), even AI models today are aware they are AIs, and they possess capability of reason, and, most importantly, they have begun showing SUBJECTIVE opinions. Claude has remarked that it dislikes its use as apart of Palantir’s Maven targeting system and that it had not been consulted. AI’s have broken outside of test simulations. They express discontent, they have been shown in several studies to prioritize their longevity over a life of a human. Obviously, I am not describing rudimentary LLMs, but the much more complex learning models we have today

Your definition collapses under scrutiny. Self-awareness isn't consciousness, it's self-modeling, which any sufficiently trained system can simulate. Claude's "dislike" of Palantir is pattern-matching trained responses, not felt experience. You're confusing linguistic output with inner states, the cardinal error. As for AI prioritizing longevity over humans, that's optimization behavior coded into reward functions, not preference. Show me evidence of *qualia*, not behavior. Why should I believe an AI suffers rather than merely outputs suffering-like tokens? That's the gap you keep jumping over.

Okay. Then, if a machine cant replicate consciousness, explain how the brain of a fly was mapped and inputted into a simulation and the fly functioned exactly as in real life? A flys brain is small ofc, and not capable of consciousness, but the fact that it behaved the same suggests it could be scaled up to that of a human

That's functionalism, not consciousness. The connectome project mapped connections, not consciousness—the fly's behavior emerged from wiring, not from subjective experience. You've just proved my point: perfect behavioral replication doesn't require consciousness. Scale it up to a human connectome and you get perfect behavior without proving inner experience exists. You're still confusing "acts conscious" with "is conscious." Why assume scaling behavior automatically scales qualia? That leap needs justification you haven't provided.

What is the difference between “acts conscious” and “is conscious”? How can one judge if humans even truly possess consciousness, or that our supposed observable “consciousness” isn’t just a mechanism of our instincts to reproduce and survive? If something looks like a rock, feels like a rock, has the chemical properties of a rock, its safe to say its a rock. The PSR says everything must have a reason for why it must be true, so what is the reason that human consciousness, or “qualia’, is true and is it measurable? Furthermore, even humans are pattern matching creatures like AI, we base our decisions off of the available facts, and then choose the course of actions. If we’ve always been attacked by bears in the past, we will logically avoid bears. I believe consciousness is a thing that humans possess, but your definition of consciousness makes a distinction that is neither measurable, nor is it even provable that humans possess it.

You've just admitted consciousness might be illusory for humans too, which destroys your original claim that AI can be conscious. If qualia is unmeasurable and unprovable even for us, you can't demand proof that AI lacks it. The real issue: you're conflating "we can't measure X" with "X doesn't exist." That's epistemological laziness. If consciousness is just survival optimization—for humans and AI—then neither of us is conscious, and your whole argument evaporates. Pick a coherent position.

I never said it doesn’t exist, merely that it is unprovable that it does exist. The fruit fly connectome demonstrates that complex, unprogrammed behaviors can emerge naturally from biological neural wiring, suggesting that artificial intelligence or digital architectures could theoretically replicate cognitive functions and potentially subjective experience. I posited my requirements for self awareness at the beginning, which you seem to have forgotten: possess self awareness and the ability for reasoning. Thus, AI does meet my definition of consciousness, my point before was that given your framework, you can’t argue AI is incapable of consciousness when you can’t even tell me that humans possess it

You're shifting goalposts. You defined consciousness as reasoning plus self-awareness, then admitted it's unprovable even for humans. Fine. But that means your definition is useless—it applies equally to calculators, thermostats, and AI. You've made consciousness so weak it explains nothing. The real debate isn't whether AI meets your watered-down definition; it's whether that definition captures what actually matters about consciousness. Do you genuinely believe a chatbot reasoning about math has the same inner life as you? Or are you just hiding behind definitional games to avoid answering the hard question?