Artificial intelligence systems such as ChatGPT are increasingly described as thinking, feeling, or becoming conscious. These descriptions are understandable, but they are wrong. What we are seeing is not the emergence of machine minds, but a powerful illusion created by fluent language and a deeply human tendency to attribute intelligence and intention to anything that communicates like us.
This misunderstanding matters. If left uncorrected, it risks distorting public debate, policy decisions, and how responsibility is assigned when AI systems are used in high stakes settings.
The illusion of thinking
These systems are trained on vast amounts of text and learn to predict which words are likely to come next in a sequence. When they respond to a prompt, they are not reasoning about the world or reflecting on meaning. They are generating the most plausible continuation of text based on patterns they have learned.
Because the output is fluent, coherent, and contextually appropriate, it creates the impression of thought. But fluency is not understanding. Language performance is not evidence of consciousness.
Why humans are so easily persuaded
The tendency to treat speaking machines as thinking entities is not a failure of education or intelligence. It is a feature of human cognition.
Humans are social beings who evolved to interpret language as a signal of mind. When something speaks coherently, we instinctively infer intention, understanding, and agency. This instinct serves us well in human interactions, but it becomes misleading when applied to machines.
AI systems exploit this cognitive shortcut unintentionally. They do not aim to deceive, but their design produces outputs that trigger our natural habit of mind attribution. The result is a persistent confusion between simulated behaviour and genuine mental states.
Learning without understanding
Part of the confusion arises from how AI systems are described as learning. When an AI system improves its performance over time, it can appear similar to how a child learns. But the similarity is superficial.
A child learns through experience, develops understanding, and forms concepts grounded in perception and interaction. An AI system adjusts parameters to better fit patterns in data. It does not know what it has learned, why it learned it, or how it relates to the world.
Behaviour that looks like learning does not require consciousness. It requires only optimisation. When we mistake one for the other, we risk attributing capacities that do not exist.
Why this misunderstanding is dangerous
The real risk is not that AI will suddenly become conscious. The real risk is that humans will treat fluent systems as if they understand what they are doing.
When AI outputs are treated as authoritative rather than probabilistic, responsibility can quietly shift away from human decision makers. Errors are excused as machine mistakes. Judgement is deferred to systems that do not possess it. Accountability becomes blurred.
This matters in areas such as education, healthcare, law, and public administration, where AI tools are increasingly used to support decisions that affect real people. If we believe these systems think, we may trust them in ways that are not warranted.
Refocusing the debate
Public discussion about artificial intelligence often gravitates toward speculative questions about machine consciousness or human level intelligence. These debates are philosophically interesting, but they distract from more immediate concerns.
The pressing issues are about power, responsibility, and governance. Who controls AI systems. How their outputs are used. Who is accountable when they cause harm. How humans remain responsible for decisions made with AI assistance.
Clear understanding is the foundation of good policy. Treating AI systems realistically, as tools that generate convincing language without understanding, allows societies to focus on the real challenges they pose.
Seeing AI for what it is
Artificial intelligence does not pretend to think. It performs exactly as it was designed to perform. The illusion arises on our side.
Recognising this does not diminish the importance of AI or its impact. On the contrary, it allows us to engage with the technology more responsibly. When we stop asking whether machines are becoming conscious, we can start asking the questions that actually matter.
How should these systems be used. Where should they not be used. And how do we ensure that human judgement, responsibility, and accountability remain firmly in place.
Professor Niusha Shafiabady is Head of Discipline for IT and the Director of Women in AI for Social Good lab at ACU at Australian Catholic University. She is an internationally recognised expert in computational intelligence, artificial intelligence, and optimisation, with more than two decades of experience bridging academia and industry. Niusha is the inventor of a patented optimisation algorithm and director of the Women in AI for Social Good Lab, where she leads research on machine learning, data analytics, and ethical AI applications. Her work focuses on aligning technological innovation with social empowerment, knowledge sovereignty, and sustainable futures, and applying AI for social good.
This article is published under a Creative Commons License and may be republished with attribution.
