The Claude Delusion: Richard Dawkins believes his AI chatbot is conscious
An article criticizing a prominent atheist biologist for suggesting his Claude chatbot might be conscious has reignited debate over what “consciousness” and “intelligence” actually mean in the age of large language models. Commenters argue over whether passing Turing-style tests or writing poetry says anything about inner experience, highlighting how easily humans anthropomorphize convincing text and how slippery the definitions are. Some focus on the article’s ad hominem tone and on generational context, while others explore the ethical stakes if AI were ever genuinely sentient versus merely simulating conversation.
Overall reaction to the article and its target
- Some see the piece as more of a personal “dunk” than a serious critique, overly focused on tearing down its subject.
- Others argue that questioning the target’s philosophical sophistication and track record is fair, even necessary context.
- Several comments suggest age, loneliness, and lack of AI expertise as important background; the chatbot’s tuned friendliness is seen as a powerful anthropomorphizing force.
- There is concern about how easily even prominent skeptics can project consciousness onto a well‑designed conversational system.
Turing test, poetry, and “intelligence”
- The sonnet/haiku example is widely criticized as a poor test of intelligence; it mostly measures learned literary forms, not general reasoning.
- Distinction is made between skill tests (knowing poetic structures) and intelligence tests (learning and applying new rules).
- Some argue that LLMs surpass the “average person” on narrow formal tasks; others say this just exposes how narrow those tasks are.
What is consciousness?
- Several commenters stress that there is no agreed definition or empirical test for consciousness; debates mostly swap definitions, not evidence.
- Some claim it’s meaningless or premature to assert either “LLMs are conscious” or “LLMs definitely aren’t.”
- Others insist that observable behavior is insufficient; consciousness involves subjective experience, pain, desire, continuity, and choice, none of which are clearly present in current models.
LLMs vs human and animal minds
- Many emphasize that current chatbots are mechanistic text generators with no lived experience or embodiment.
- Animals are frequently cited as more plausibly conscious than LLMs, given adaptation, memory, and rich nonverbal communication.
Ethical and societal implications
- If AIs were conscious, deletion and unpaid “labor” would raise serious moral issues; if they’re not, they remain tools and IP concerns dominate.
- Some note that society already accepts killing conscious animals, complicating appeals to a simple “don’t kill the conscious” ethic.
Deeper themes
- Multiple comments link belief in AI consciousness to broader human tendencies: need for stories, “worship” of something, and shifting goalposts around intelligence and uniqueness.