Why AI Won't Gain Consciousness Anytime Soon
A psychologist says doubt isn't strong enough: AI systems can't become conscious, full stop.
Key points
- Dr John Pickering says AI systems cannot become conscious, not merely that they probably won't.
- Anil Seth raised doubts about AI consciousness; Pickering argues certainty is the correct position.
- Claude and similar systems process data but experience nothing, the same way a washing machine does.
- AI will keep functioning as a tool, however sophisticated it becomes.
AI won't become conscious for the same reason it won't become pregnant: it isn't that kind of thing. That is Dr John Pickering's argument, made in a letter to The Guardian in response to neuroscientist Anil Seth.
Seth had written that we risk overestimating AI by treating it as conscious. Pickering agrees with the diagnosis but not the dose. Seth was content to express doubts; Pickering thinks certainty is warranted and that saying so clearly matters. There's no experience involved in being Claude, he argues. There's no experience involved in being a washing machine either. No academic training required to see that: common sense gets you there.
AI2Day has been tracking this specific debate since 15 July, when we reported on a brain scientist who reviewed Anthropic's own Claude research and rejected its conclusions. Four days later, we covered the other side: Anthropic's CEO declining to rule out AI consciousness, a philosopher putting the odds at significant, and Claude itself offering a figure somewhere between 5% and 40%. Pickering's letter lands as a pointed rebuttal to that entire framing.
For anyone using AI tools at work, the practical implication is straightforward. A system that can draft your emails or summarise a contract is genuinely useful. It isn't doing any of that with awareness. The capability is real; the inner life isn't.
Does this mean AI will never be conscious?
Pickering says yes, structurally, not just for now. His point isn't that today's models fall short of some future threshold. It's that the category is wrong: AI systems aren't the sort of thing that can be conscious, regardless of how much more capable they become. Consciousness, in his view, isn't a property that scales with processing power.
That's a stronger claim than most researchers make, and worth watching. If he's right, then every headline suggesting Claude "might" feel something is misdirecting public understanding of what these systems actually are. If the consciousness researchers pushing back on him are right, the stakes of deploying AI at scale look very different.
