As artificial intelligence becomes increasingly capable of conversation, reasoning and problem-solving, researchers are examining a question once largely confined to science fiction: could a machine ever have a genuine inner experience, including sensations such as pain or pleasure?
The issue remains unresolved, partly because scientists still do not fully understand why humans and other animals are conscious. Researchers are now exploring whether consciousness depends fundamentally on biological tissue or instead arises from particular forms of information processing that could, in theory, be reproduced artificially.
Could an AI System Have a Subjective Experience?
Ulysse Klatzmann, a postdoctoral researcher in cognitive and computational neuroscience at Sainte-Justine hospital and Université de Montréal, and Adrien Doerig, a professor at Freie Universität Berlin, have examined the question from a biological perspective.
Their article, What biology can, and cannot, tell us about conscious AI, does not attempt to determine whether today’s AI systems are conscious. Instead, it asks a more fundamental question: what can biology tell scientists about whether artificial consciousness is possible at all?
Defining What Consciousness Actually Means
A major difficulty is defining consciousness itself.
The term can refer to being awake, perceiving the surrounding environment or engaging in self-reflective thought. Klatzmann and Doerig focus instead on phenomenal consciousness — the existence of subjective experiences.
Seeing a blue sky, hearing music or experiencing pain are examples. Humans know these experiences occur because they directly experience them, but scientists still do not fully understand how the brain produces them.
“We humans share an intuition about what consciousness is,” Klatzmann said. “We spontaneously attribute consciousness to humans and other animals, but not to a bench or a machine.”
AI is challenging that intuition. Modern systems can generate coherent text, solve mathematical problems, recognise images and sustain sophisticated conversations, sometimes creating the impression of interacting with an entity that understands what is being said.
But behaviour alone raises a difficult question: if a machine acts as though it is conscious, does that mean it actually experiences anything?
“We have long linked consciousness to intelligence and cognition,” Klatzmann observed.
AI can now reproduce aspects of those cognitive abilities without providing evidence that it possesses subjective experiences. Distinguishing between simulated consciousness and genuine experience therefore remains a central scientific challenge.
Does Consciousness Require Biology?
One prominent position in the debate is biological naturalism, which proposes that biology is indispensable to consciousness. Under this view, living systems possess characteristics necessary for subjective experience that artificial systems do not.
Every conscious entity currently known to science is a living organism. Humans and other animals have biological brains, while computers do not.
However, Klatzmann and Doerig argue that this observation alone does not demonstrate that biological material is essential. Establishing that conclusion would require identifying precisely what biology contributes to consciousness.
The Brain-Replacement Thought Experiment
The researchers illustrate the problem with a thought experiment.
Imagine replacing part of a person’s brain with a computer chip capable of performing exactly the same functions. The individual continues to speak, see, respond and behave exactly as before.
Would their conscious experience remain unchanged?
A strict biological-naturalist interpretation would suggest that consciousness had been altered because part of the biological brain had been replaced.
Yet if behaviour, responses and self-reports remained identical, there might be no observable way to distinguish the person’s condition before and after the replacement.
The researchers argue this presents an important experimental difficulty. If a theory links consciousness exclusively to biology but produces no testable predictions, verifying it scientifically becomes challenging.
What the Brain Does May Matter More Than Its Material
An alternative approach is functionalism, which proposes that consciousness depends on what a system does rather than exclusively on what it is made from.
Under this view, the brain could generate consciousness because it carries out particular forms of information processing. Biology might therefore be important without necessarily being an absolute requirement.
Potentially significant characteristics include the continuous dynamics of neural activity, the self-regulating nature of living organisms, the way neurons integrate information and interactions between different levels of brain activity.
The human brain operates differently from conventional digital computers in several respects, but those differences do not automatically establish which properties produce consciousness.
For example, neural processes are continuous and dynamic, while computers typically operate through discrete digital processes. Weather systems are also continuous and highly complex, yet computers can model aspects of them effectively.
Researchers would therefore need to demonstrate that a particular brain function directly contributes to consciousness before concluding that it is indispensable.
Scientists Warn Against Relying on Intuition
Klatzmann and Doerig also caution against arguments based primarily on intuition.
One argument holds that because every known conscious being is biological, consciousness itself must require biology. The researchers point out that the same reasoning could incorrectly suggest that humans must live on Earth simply because every known human has done so.
Another argument describes large language models as systems that are “just trained to predict the next word”, implying they cannot possess genuine cognitive abilities.
The researchers argue that reducing a complicated system to one underlying process does not necessarily describe all the properties that may emerge from it. Similarly, humans could be characterised as organisms “just selected to pass on their genes”, without that description capturing the full range of human abilities.
“The history of science shows that things that seem self-evident may turn out to be false. Matter seems solid, but we now know it is mostly empty space. Time seems universal, but relativity shows it isn’t. Why would our intuitions about consciousness be infallible?” Klatzmann said.
Could Future Artificial Intelligence Become Conscious?
The researchers do not offer a definitive answer about whether future AI could develop consciousness.
Klatzmann describes himself as agnostic on the possibility.
“I would be surprised if current AI models develop consciousness,” he said, “but we can’t entirely rule it out.”
Future technological developments could further complicate the debate. Even if biological properties ultimately prove essential for consciousness, humans may eventually develop artificial systems incorporating some biological characteristics.
Alternatively, if consciousness depends on particular functions rather than biological material itself, scientists will need to identify which properties of the brain are essential to subjective experience and determine whether artificial systems can reproduce them.
For now, the possibility of conscious AI remains an open scientific question. Resolving it may depend less on how convincingly machines imitate human behaviour and more on achieving a deeper understanding of how consciousness emerges in the human brain itself.
Cory Weinberg is a contributor to Sproutwired.com, covering a wide range of topics including news, politics, business, technology, sport, entertainment and lifestyle. He focuses on delivering clear, balanced reporting that helps readers stay informed about current events and emerging developments. Cory’s work highlights relevant stories, practical insights and important issues affecting communities and industries, with an emphasis on accuracy, clarity and information that readers can trust.
