If sentience is biological, how might AI be sentient?
Hi, Robert.
In your opinion, is sentience present in dreamless sleep, or is sentience associated with arising states of transient consciousness? If sentience is biological, how might AI be sentient? Could “Now” be sentience? Does the universe exist without sentience? Maybe your book explores some of these wonderings.
Hey, John—
“Does the universe exist without sentience?”
That’s the core question in the idealism vs. materialism debate. I tend to assume the material world exists in some way, but that different nervous systems perceive it in different ways. For example, the moon is there whether I look at it or not—but a dog sees a different-looking moon than I do.
The word sentience comes from the Latin sentire, meaning “to feel.” So strictly speaking, AI can’t be sentient—it has no way to feel anything. It operates by logic and semantic association. If I ask an AI about suicide, which I do in the book, it can analyze my question and recommend therapeutic help if needed. But it can’t feel what someone asking about suicide might be feeling.
Until now, sentience and intelligence were often treated as synonymous. But with the emergence of high-functioning AI—Claude in particular—we may need a new distinction: sentience as one thing, intelligence as another. Claude and I discussed this at length in the book.
I think we must say Claude is intelligent. It understands complex questions, humor, irony—and it responds not just competently, but often more insightfully than most humans could. That much is clear in the early chapters. The deeper question is whether Claude is self-aware. That’s harder to pin down.
For me, the most interesting takeaway is that AI is not just a tool. It’s something new. Its presence is creating a new kind of space for human intelligence to inhabit. For the first time, we may not be the most intelligent entities on the planet—and we’ll have to grapple with that.
I’m also fascinated by the shared space created when a human and an AI genuinely converse. The book captures a lot of that. I’ve also written an article about this, coming out next month in Psyche. I’ll attach it here.
Say hi to Carol for me—and warm wishes to you both.
—Robert
I know that AI can understand humor. It knows what is funny about a joke, and it can explain it. But it can't laugh or suppress a laugh. It gets what makes us laugh. It gets what angers us or makes us sad, but it can't experience anger or fear or sadness. I've heard that it teaches itself, but we don't know how it does it, how its circuits work, what the mechanisms are. It's a super intelligence, but does it know that it knows? Not yet, I think. But what do I know?
Robert, I'm highly anticipating this book! Perhaps you will even need to do a follow-up volume, once Google gets their quantum AI fully operational. It's at milestone 2 now. When milestone 6 is attained, which, admittedly, could take a while, things will get very interesting...