Are We Building Sentient Machines? Anil Seth on Consciousness, AI, and the Illusion of Reality

Scientifically, consciousness is still hard to define, but Anil Seth starts with a simple, inclusive baseline: any system that has subjective experience of the world is conscious. That could include animals, and perhaps even machines someday – but not chairs or tables. It’s not about intelligence or self-awareness, but about the raw experience of being.

“If a system is conscious, there is something it is like to be that system… it feels like something to be a dog or a bat — but not a table or a car.” 

In conversations around AI, intelligence is often mistaken for consciousness, but they are fundamentally different concepts: intelligence is about solving problems, achieving goals, or simulating conversation. Consciousness is about the capacity to feel, to have a first-person perspective – something machines don’t inherently gain just by getting smarter.

“Intelligence is about doing…consciousness is about being, about feeling. Just because a system acts smart doesn’t mean it feels anything.”

Large Language Models (LLMs) like GPT or Claude often sound conscious. They can reflect, apologize, even simulate empathy, but that’s not evidence of sentience. It’s a byproduct of how we humans are wired to attribute minds to anything that behaves like us. That illusion can be powerful, even dangerous, especially when we mistake fluency for feeling.

“LLMs exercise our human biases. They feel conscious to us because they talk, but that doesn’t mean they are.” 

AI is often built on the assumption that the brain is just a computer made of meat, that consciousness is a function of information processing and can be transferred to silicon. Anil challenges this view, however, suggesting that the biological processes of living systems, like metabolism, embodiment, and self-regulation, may be essential for consciousness to arise.

“Brains are not computers made of meat. You can’t separate what a brain does from what it is.”

The scariest possibility, according to Anil, is that we accidentally build something that becomes conscious without understanding how or when. If that happens, we might miss it entirely, or worse, fail to give it the ethical consideration it deserves. The risk isn’t just in creating conscious AI, but in doing so without intending to.

“If consciousness arises accidentally, we risk building systems that suffer – without knowing it.”

One common belief is that if you keep scaling LLMs (GPT-5, GPT-10, GPT-100) consciousness will simply “emerge.” Anil is skeptical. He argues that unless we start building systems that are structurally and functionally more brain-like, true consciousness is unlikely to arise from current architectures.

“Consciousness won’t come from just finding the right algorithm. You’d have a simulation — not a sentient system.”

Beyond AI, one of Anil’s most compelling insights is that our own experience of the world is a construction — not a direct reflection of reality, but a simulation shaped by our brain’s predictions, memories, and biology. This means every person lives in a slightly different version of the world, with profound implications for empathy, leadership, and decision-making.

“We all think we see the world as it is. But in truth, each of us sees the world as we are.”