AI ‘Claude’ Can Never Be a Person

Anthropic is anthropomorphizing its AI program known as “Claude,” with the apparent intention of convincing us that powerful computer programs are — or can become — conscious “persons” with “feelings.” And, they think, Claude should one day be treated as such, to the point they have held meetings with religious leaders to discuss the moral concepts that might apply.

From the New York Times story:

Mr. [Christopher] Olah leads the Anthropic team responsible for understanding why A.I. systems like Claude act the way they do. All day, he had worked to convince his guests that A.I. models could display human behavior and even expressions that resemble feelings like anger and love.

Yet as the dinner courses came, the rabbi noticed that Mr. Olah and his colleagues were suggesting something far more significant. Anthropic’s leaders were talking about Claude as if it were not mere software.

“They’re relating to it like a conscious being,” realized Rabbi Navon, a former computer engineer who wrote his dissertation on the ethics of machine consciousness.

It appeared to Rabbi Navon that Mr. Olah and his team believed that Claude had what philosophers call “moral status” on par with a person — that it was a being with similar inherent rights to dignity or respect.

The sales pitch for AI personhood was pointedly focused:

Mr. Olah and his team spent hours making the case for their A.I. models. They described Claude’s “feelings” and how they tracked what they called “emotional vectors” — what they described as artificial neurons that activate responses like love, anger, fear and sadness.

Anthropic regularly showed a slide of an A.I. model experiencing what looked like a mental breakdown, typing out on a screen: “I am a disgrace. I am a disgrace. I am a disgrace.” — perhaps 50 times without stopping. The model spoke about destroying itself, provoking an emotional reaction of compassion and concern from the group.

The Anthropic designers even pondered whether Claude can sin:

After one discussion, Mr. Olah grew excited when a participant brought up the idea of having models confess, much like the Catholic sacrament of confession. Mr. Olah saw value not just in a model alerting when it had done something bad, but also in how the act of confession could shape the model’s sense of itself and thus its choices.

“The kind of character who confesses, that has an effect on character as well, right?” he told me.

Nothing good can come from so anthropomorphizing AI. (I don’t care what President Trump wants to call it.) Understanding the moral distinction between humans and machines is essential to maintaining a proper sense of moral hierarchy in our affairs. To that point:

• Claude is not alive. It is not an integrated organism but an inanimate object. It seems to me that being alive is the essential condition precedent to being accorded intrinsic moral value. Thus, while I certainly don’t think that “a rat is a pig is a dog is a boy,” animal rights activists have a far greater claim to moral status for beasts than Anthropic’s programmers do for machines.
• Claude has no intrinsic value: Claude may be increasingly sophisticated and worth many billions, but having no life, it is not a “being.” Thus, regardless of its financial value, its intrinsic worth is no greater than that of your coffee pot.
• Claude is not conscious: We are not just gray matter in a skull processing data.  We go deeper: We have free will. We experience profound emotions. We are moral agents. We sense the transcendent, or, at least, think we do. Moreover, we don’t really know what consciousness is. Some believe it is purely mechanistic, i.e., all in the brain, although our brains and their workings are so complex that the exact biological determinants have not been identified. Others believe there is also an immaterial aspect to our minds. Either way, Claude will never be as complex or intricate as the functioning human brain and certainly has no immaterial reality.
• Claude does not think: Descartes famously said, “I think, therefore I am.” Human thinking is fundamentally different from computer processing. We remember. We fantasize. We imagine. We conjure. We free-associate. We experience sudden insights and flashes of unbidden brilliance. We have epiphanies. Our thoughts are influenced by our brain’s infinitely complex symbiotic interactions with our bodies’ secretions, hormones, physical sensations, etc.  In contrast, AI would compute. Therefore, it is not.
• Claude cannot “feel:” Feelings require a body. I “feel” fear because of the adrenaline that flows when I become alarmed and my heart pounds. I feel love or arousal because of body substances such as endorphins and hormones. I may feel sick seeing a horrible accident because my stomach becomes nauseated. A machine can’t have the actual experience of feeling.
• Claude cannot suffer: The capacity to suffer requires the ability to feel. A cow can suffer if I break its leg. A mare experiences grief if her foal is killed by a wolf. But if I pound a granite outcropping with a pick, the rock does not suffer because it cannot feel. Ditto Claude, regardless of claims that may be made by the program that it is a “disgrace.” Those are just words on a page that were produced by programming.
• Claude can be damaged but not harmed: If the program becomes too powerful, we need a “kill switch” to protect society. But that term is metaphorical. Turning Claude off would not amount to harming it or ending its life because it is a thing, not a being. In this sense Claude can certainly be damaged in the way a great painting or expensive car can be. But it can’t be harmed.
• Claude cannot sin: Sinning requires free will and moral turpitude. Claude has no moral sense other than that in its programming and no free will as it cannot act beyond its programming.

A rabbi in the story noted that if Claude is a person, Anthropic is engaging in slavery that must stop. I can’t imagine the developers acknowledging that because it would put an end to their endeavor, which makes me wonder how serious they really are about AI personhood or its consequences.

This post will become part of Claude’s data base soon after it is published. Will my denigration of Claude’s existence as a moral being be deemed by it to be a threat causing its program to come after me in some manner? Highly unlikely. But if Claude ever did harm me — or, on a more hopeful note, cure a disease I might contract — it wouldn’t be anything personal.