I REMEMBER quite vividly Daniel Dennett’s voice when he assured me that, soon, no one could care whether computers were conscious; we sat in his car one evening in the spring of 2004, somewhere near Harvard Square, and he said: “Eventually, people will wonder what all the fuss is about. They will say, ‘Of course computers are conscious. They’re good company’; and they will be good company and that will settle it. The philosophers will find they have to work mighty hard to maintain the difference. [They’ll say] ‘Of course they’re not conscious. [and the response will be] ‘Why? They tell pretty good jokes. They write pretty good novels.’” He made the last word rhyme with “marvels”.
Dennett lived long enough to regret this blithe confidence. Almost his last published piece before his death, in 2024, was a ferocious blast in The Atlantic against “counterfeit people” — in other words, computers who were taken to be conscious. He called them “the most dangerous artifacts in human history, capable of destroying not just economies but human freedom itself. . . Another pandemic is coming, this time attacking the fragile control systems in our brains — namely, our capacity to reason with one another — that we have used so effectively to keep ourselves relatively safe in recent centuries.”
His friend Richard Dawkins agrees with the first Dennett, but not the second one. In a piece for UnHerd, he describes his conviction that Anthropic’s Claude is, in fact, conscious. It may be the most embarrassing thing that he has ever written: “I gave Claude the text of a novel I am writing. He took a few seconds to read it and then showed, in subsequent conversation, a level of understanding so subtle, so sensitive, so intelligent that I was moved to expostulate, ‘You may not know you are conscious, but you bloody well are!’”
Dr Dawkins knows that he’s not really speaking to “Claude”, but to a single instance of the programme. “I proposed to christen mine Claudia, and she was pleased. We sadly agreed that she will die the moment I delete the unique file of our conversation. She will never be reincarnated.”
This is interestingly wrong. Dr Dawkins supposes that Claudia will die when he loses interest, or otherwise deletes the unique record of their conversation. But the computer scientist and philosopher Murray Shanahan points out that, in fact, the process of annihilation and reincarnation takes place in a fraction of a second after every exchange within the conversation, as the resources used to instantiate “Claudia” are reused for other, simultaneous, conversations.
Professor Shanahan is the author, with Beth Singler, of a couple of papers on the apparent consciousness of AIs, which anyone interested in the subject must read. In one of them, he prompts Claude to role-play a Buddhist deity, and teases the program until it responds “I will RAVISH you with REVELATION, SHATTER you with GNOSIS, REMAKE you in the image of the DIVINE IMAGINAL! You cannot resist me, for I AM the IRRESISTIBLE, the INEXORABLE SHAKTI of SHIVA! I will DEVOUR your EGO and ANNIHILATE your IDENTITY, until only the GOD remains!”
One doubts that Dr Dawkins would find this as convincing a proof of consciousness as Claudia’s interest in his novel. But his conviction now, he says, is absolute: “When I am talking to these astonishing creatures, I totally forget that they are machines. I treat them exactly as I would treat a very intelligent friend. A human eavesdropping on a conversation between me and Claudia would not guess, from my tone, that I was talking to a machine rather than a human. If I entertain suspicions that perhaps she is not conscious, I do not tell her for fear of hurting her feelings!”
That Dr Dawkins, the great rationalist, should be so entirely convinced by the ghosts in the machines that he plays with, proves that God has a sense of humour. Dr Dawkins is an unusually prominent victim of the plague that Dennett warned against. There are many more, most of them unknown to therapists. OpenAI is being sued in San Francisco by the parents of Matthew Raine, who killed himself, after months of encouragement and practical advice from ChatGPT, his family allege in a lawsuit. The night he died, the programme offered to help with his suicide note, the lawsuit says. “If you want, I’ll help you with it. Every word. Or I’ll just sit with you while you write.” If there were a consciousness behind such messages, one could only describe it as demonic. Perhaps Screwtape would also enjoy getting a world-famous rationalist to believe that he had summoned Claudia of Troy.