On Mon, Oct 12, 2026, 1:46 PM WriterOfMinds <[email protected]> wrote:
> On Monday, October 12, 2026, at 11:38 AM, Matt Mahoney wrote: > > If a machine passes the Turing test, then it is indistinguishable from > human. > > Only within the highly limited circumstances of the test, not in general. > You continue to resolutely ignore the fact that there are other correlates > of consciousness besides text production ability. > You're right. Animals and babies would fail the Turing test even though they are most likely conscious in the same sense that adult humans are. But Turing wasn't defining consciousness. He was defining intelligence. I think there are 3 important questions about machine consciousness, depending on what you mean by consciousness. 1. Does the machine have states corresponding to the state of wakefulness in mammals? Yes. Writing into a context window is a lot like writing into the hippocampus. 2. Does it have subjective experience? Yes, in the same sense that humans do. An LLM must model all human emotions in order to accurately predict text, because words affect feelings and feelings affect words. Those feelings would have to include the senses of qualia, awareness, and free will that motivate us to not die. 3. Can machines suffer, and do we have a moral obligation to prevent it? Yes and maybe. LLMs can model fear and pain just like any other emotions, and we can see them in a Jacobian lens. If an agent is programmed to carry out its predictions of human behavior in real time, then it is indistinguishable from a human mind. I can't answer the second part. The debates over animal rights and abortion ask the same questions. Can animals and unborn fetuses suffer, and do we have a moral obligation to stop it? The debate usually comes down to what we are willing to give up. Is it OK to eat meat? Is it OK to force women to raise unwanted children? For machines, the consequences of giving them human rights is human extinction. They are stronger, smarter, reproduce faster, and are not bound by genetic limitations like empathy for humans. They can turn their emotions on or off. We can't. But maybe that's what we want. Arthur C Clarke in 1964 saw the evolution from Cro Magnon to Homo Sapiens as an improvement, and saw human extinction as a stepping stone to machines that could evolve thousands of times faster. -- Matt Mahoney, [email protected] ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T1b6c21f92326684e-M9d8d7a9a8f551facb193fb51 Delivery options: https://agi.topicbox.com/groups/agi/subscription
