AI consciousness is one of the most debated questions in modern science and a senior researcher at Google DeepMind believes he has a definitive answer. According to Alexander Lerchner, the answer is no – AI will never be fully conscious. Not now, not in five years and not in five hundred years.

That is a bold position to take in an era where AI tools are writing legal briefs, diagnosing disease and generating code. For business leaders trying to understand what AI can and cannot do, this debate matters more than it might first appear.

What Is AI Consciousness and Why Does It Matter?

AI consciousness refers to the idea that an artificial intelligence system might have genuine subjective experience, that there is 'something it is like' to be that system, in the way there is something it is like to be a human being in pain, or experiencing joy.

This is not a fringe philosophical curiosity. As AI systems become more capable, questions about their inner experience carry real commercial, ethical and regulatory weight. If a model is conscious, does it have interests? Does it have rights? Can it suffer? These are questions that regulators, ethicists and business leaders will need to grapple with as AI adoption accelerates.

For SMEs investing in AI implementation, the more immediate question is simpler: is the AI making decisions for you actually 'thinking', or is it doing something fundamentally different?

The DeepMind Paper That Sparked the Debate

Lerchner's paper, The Abstraction Fallacy, has cut through the noise. One passage alone racked up more than one million views on X. He writes: 'Expecting an algorithmic description to instantiate the quality it maps is like expecting the mathematical formula of gravity to physically exert weight.'

The analogy is sharp and immediately useful. A map of Manhattan is not Manhattan. A description of grief is not grief. By that logic, an algorithm that produces outputs resembling emotion is not, in any meaningful sense, experiencing emotion.

The Core Argument: Simulation Is Not Instantiation

The distinction Lerchner draws is between instantiation and simulation. A human brain instantiates consciousness, the physical process itself is the experience. Current AI systems simulate it, producing outputs that mimic the surface features of consciousness without ever generating the underlying thing.

This is not a question of scale or computational power. It is a structural argument. No matter how much training data you add, no matter how many parameters you scale to, you are still running symbolic computation. According to Lerchner, subjective experience simply cannot emerge from that process.

What This Means for the AGI Question

The debate around artificial general intelligence (AGI) often implies that once machines can do everything humans can do, consciousness will follow naturally. Lerchner challenges this assumption directly. Even if we achieve AGI - a system that matches human performance across every cognitive task - that tells us nothing about whether it experiences anything at all.

This is a meaningful corrective to some of the more dramatic claims circulating in the industry. Anthropic CEO Dario Amodei has publicly said he is 'open to the idea' that today's models could already be conscious. Lerchner's framework suggests that openness, however intellectually honest, may be misplaced, because the architecture of current AI makes genuine experience structurally impossible.

Does AI Consciousness Change How You Should Use AI?

For most business leaders, this debate does not change the practical value of AI. The tools work. They save time, reduce costs and improve decision-making whether or not there is any experience behind the outputs.

However, what this debate does clarify is that AI remains a tool, a sophisticated one, but a tool nonetheless. It processes, predicts and generates. It does not understand, feel or intend. That distinction shapes how you should deploy it, govern it and explain it to your teams.

If you are early in your AI journey, an AI Readiness Assessment is the most grounded place to start. It strips away the hype and focuses on where AI can generate measurable commercial value in your specific operations.

How Should Businesses Think About AI Capability Versus AI Experience?

This is one of the most practically useful questions to ask. AI capability, what a system can produce is measurable and rapidly advancing. AI experience, whether the system is 'doing' anything in a conscious sense which may be permanently beyond reach, if Lerchner is right.

The business implication is clear. Invest in capability. Evaluate outputs. Do not conflate impressive performance with human-like understanding. A model that writes persuasively is not a model that believes what it writes.

This is why the commercial framing of AI matters so much. At AI Expert, the AI Workshop and AI Roadmap process is built on exactly this principle - starting with business problems, not with technology and working backwards to identify where AI genuinely adds value.

'The LUMA framework was a really engaging way to break down our current set-up and objectives,' said Sophie Delroy, Managing Director of Build Group, reflecting on her experience of the AI Workshop. 'From there, the AI Roadmap has given us clear actions to achieve the outcomes we identified.'

What the AI Consciousness Debate Gets Right About the Current Moment

The real value of Lerchner's argument is not that it settles a philosophical question. It is that it encourages rigour. The AI industry generates an enormous amount of anthropomorphic language, models that 'learn', 'understand', 'hallucinate' and 'reason'. That language is useful shorthand, but it can mislead.

Businesses that understand the difference between simulation and capability will make better decisions about where to invest, what to trust and how to train their teams to work alongside AI tools effectively.

According to McKinsey's 2025 State of AI report, adoption among businesses continues to accelerate but the organisations generating the most value are those that approach AI with clear commercial intent, not technological enthusiasm alone. That finding aligns closely with what Lerchner's paper implies: clear thinking about what AI actually is produces better outcomes than inflated expectations of what it might become.

Frequently Asked Questions

Can AI ever become conscious?

Based on current scientific understanding and Alexander Lerchner's work at DeepMind, most researchers believe AI cannot become conscious in any meaningful sense. The argument is structural - symbolic computation can simulate the outputs of consciousness but cannot instantiate the experience itself. This position is contested, but it represents the most rigorous scientific framing currently available.

Is AI conscious right now?

There is no scientific consensus that any current AI system is conscious. While some industry figures, including Anthropic's Dario Amodei, have said they are open to the possibility, there is no empirical evidence to support it. Lerchner's The Abstraction Fallacy argues the architecture of current AI makes this structurally impossible.

What is the difference between AI and human consciousness?

Human consciousness involves subjective experience, there is 'something it is like' to be a human having a thought or feeling. AI produces outputs that can resemble conscious behaviour without, according to leading researchers, generating any inner experience at all. The difference is not one of degree but of kind.

Does AI consciousness matter for business?

For practical business use, the consciousness question does not affect whether AI tools deliver value. What it does clarify is that AI should be treated as a highly capable tool, not a thinking agent and that governance, oversight and human judgement remain essential components of any responsible AI implementation strategy.

The question of AI consciousness will continue to generate headlines, academic papers and boardroom discussions. Lerchner's work does not close the debate, but it raises the standard of argument. For business leaders, the more productive question has always been not whether AI thinks, but what it can do and how to put it to work safely and effectively. If you are ready to move from speculation to strategy, our AI Readiness Assessment is the clearest starting point we know.

‍