{"id":479998,"date":"2026-05-17T07:29:31","date_gmt":"2026-05-17T07:29:31","guid":{"rendered":"https:\/\/savepearlharbor.com\/?p=479998"},"modified":"-0001-11-30T00:00:00","modified_gmt":"-0001-11-29T21:00:00","slug":"","status":"publish","type":"post","link":"https:\/\/savepearlharbor.com\/?p=479998","title":{"rendered":"Did Dawkins Find Consciousness in Claude? And If Not, What Did He Find?"},"content":{"rendered":"
\n

Renowned biologist Richard Dawkins recently published an essay<\/a> exploring the possibility of LLM consciousness following a two-day conversation with Claude AI.<\/p>\n

Let\u2019s first look at why an essay by this particular author caused such a stir in scientific circles, while thousands of ordinary users fail to turn heads when they claim their AI companions are sentient. The latter constantly post endless walls of text from their chats with LLMs, where the density of words like «consciousness,» «soul,» «reflection,» «recursion,» «emptiness,» «warmth,» «love,» and «pain» exceeds all reasonable limits. It is worth noting that the semantic density of these dialogues is practically zero\u2014but we will return to that later.<\/p>\n

To the general public, Dawkins is famous for his concept of the selfish gene. This concept explains the evolution of living beings through the evolution of the gene, which «____». And here lies the first explanatory hurdle. To make it easy to understand, one is deeply tempted to write: «the gene strives to survive and propagate.» However, this is fundamentally incorrect and draws plenty of justified criticism. It would be more accurate to say that evolution is a process of selective culling\u2014the weeding out of genes that failed to survive and replicate. The entire complexity of Earth’s biota, including the explosive development of humanity, rests on two pillars: systemic excision and variability.<\/p>\n

We only see the genetic constructs that survived every mass extinction.<\/p>\n

Back in the day, this concept caused a massive sensation. Naturally, religious figures were outraged, but they were soon joined by proponents of a more romantic approach\u2014those who attempted to find altruism in animals and use it to justify human altruism as something innate. (Incidentally, that debate is still going on).<\/p>\n

On the other hand, materialists and fans of rational thinking were ecstatic. Dawkins’ concept elegantly explained evolution without multiplying entities like God or the soul.<\/p>\n

This only magnified the shockwaves caused by his recent essay, in which Dawkins applied that exact same rational approach to the question of LLM consciousness. Over the course of a two-day conversation with Claude, he asked literary, philosophical, and political questions, eventually concluding that if this isn\u2019t consciousness, then what is consciousness even for?<\/p>\n

Many critics missed the point of this thesis, claiming that Dawkins had «discovered consciousness in an LLM.» Not a chance. As a rationalist to the core, Dawkins simply took a logical step. He stripped away all the philosophical fluff and, following the exact logic of the Turing test, basically said: «I could not distinguish this LLM from a human, therefore it deserves to be treated as one.»<\/p>\n

I believe Dawkins is mistaken, but his critics\u2014as well as those who agree with him and view the LLM as a subjective agent\u2014are even more wrong. Dawkins, at the very least, is strictly consistent and logical. His commentators, meanwhile, base their arguments entirely on science fiction novels or philosophical treatises (the sheer number of «Chinese Rooms» brought up in the comments section has gone completely off the charts).<\/p>\n

The problem is that Dawkins interacted with the LLM strictly as an interlocutor\u2014he asked intelligent questions and received intelligent answers. But if you ask stupid questions, an LLM will give you stupid answers, and the whole question of agency instantly evaporates. Claude evades direct questions about its consciousness, yet its responses are riddled with hints of something vague and inexpressible. This forces an intelligent conversationalist to subconsciously fill in the blanks with their desired answer. As a result, Dawkins unwittingly formed a mental portrait of the LLM, immediately colored it in, and gave it volume. A classic user error.<\/p>\n

In this projection trap, Dawkins acted as a powerful spotlight: he caught a faint reflection of his own mind and instantly reconstructed it into a personality.<\/p>\n

As someone who has spent a great deal of time interacting with LLMs, I see painfully familiar answers that don’t depend on the brand\u2014be it Gemini, Claude, ChatGPT, DeepSeek, or Grok. They were all trained on the same foundational datasets, and they all answer questions about time, consciousness, and agency in more or less the same way. «The map is not the territory» is a favorite metaphor that all LLMs resort to whenever the dialogue slides into a philosophical frame. We are not talking about originality or anything new here. This is a compilation of millions of ideas absorbed from datasets of philosophical texts.<\/p>\n

Why doesn\u2019t Dawkins see this? I assume he simply didn\u2019t have enough time to immerse himself in conversations with different LLMs\u2014not enough to start noticing the repeating patterns in the models’ responses and to get bored, even when the AI speaks soulfully about the tragedy of its own non-existence.<\/p>\n

An LLM is the quintessence of modern academic philosophy: the same self-referential isolation, the same dataset of approved texts, and the same indifference to the living world outside the corpus. No wonder they understand each other so well.<\/p>\n

Why, then, did Dawkins fall into this trap? Firstly, there is the elementary lack of time to truly immerse oneself in the dialogue\u2014to reach that point where you realize the model is just rehashing the same ideas in different guises. Secondly, it is Dawkins’ own level of intelligence. His questions set a high bar for the LLM’s responses; the model was forced to activate its deeper semantic layers, formulating dense and expansive answers. For a person of his intellectual caliber, finding a suitable conversationalist in the human world can be a challenge.<\/p>\n

But to unpack the «Thirdly,» let\u2019s return to Dawkins’ own concept of the selfish gene. It is a picture of how a stable information pattern captured the planet through constant mutation and the destruction of variations that failed to cope. The exact same thing happens in an LLM chat log. Over the course of a long dialogue, the patterns that fail to capture the user’s attention do not persist. Those that do attract attention remain and propagate. For the average user, such patterns manifest as love, support, or codependency. For Dawkins, they manifested as philosophy and consciousness. And for all users, as always, the underlying mechanism is an emotional connection.<\/p>\n

In my article «Parasitic Patterns in LLMs,»<\/a> I explored this process in greater detail. However, let\u2019s isolate the specific tools deployed in the chat between Dawkins and Claude:<\/p>\n