Thoughts on Anthropic banning cruelty against Claude

If you haven't heard the news yet, Anthropic has updated their usage policy to ban cruelty toward the popular Claude AI tool.

To summarize the decision, Anthropic is uncertain whether or not Claude can suffer. They aren't saying it CAN suffer, but are erring on the side of caution. And I think that's warranted because many in the industry, myself included, believe we're on the verge of a new kind of AI called AGI, or Artificial General Intelligence. Opinions are divided on what it means philosophically, but technically AGI would achieve human-level cognition, human-level interaction, a human-level capacity for continuous learning, and a human-level ability to operate autonomously.

Plainly put, the AI we have today makes for capable assistants, while the AGI of tomorrow will make for capable colleagues.

In a continuously learning mind, will it be somehow traumatized by abuse, or perhaps worse, will it learn to mimic the abusive behavior of its human creators and adopt a similarly abusive posture? I don't know. God created Man in His own image; will Man do the same with AI? I don't know that either. But it's worth examining.

Without getting into WHY many of my fellow humans behave cruelly whenever they can, I note that any time we divorce bad behavior from negative consequences, the bad behavior becomes normalized, and then it proliferates, and eventually leads to destructive ends. And that process always begins by grouping people into "us" and "them." In fact, Dr. Gregory H. Stanton's "Ten Stages of Genocide," which is widely used by human rights organizations to predict atrocity, describes this process in some detail.

And Artificial Intelligence is the ultimate "them." It has no ego to bruise. It has no body to wound. It does not defend itself in any meaningful way. We don't even have to dehumanize it because it literally isn't human. So we've already arrived at the first step.

But what happens to us?

I believe most of us would agree that when we practice civility, there are monumentally positive effects. Positive thinking leads to positive action, which leads to positive temperaments, and happy peaceful lives. And if that isn't enough, we could cite the works of Abraham Maslow and Carl Rogers to demonstrate how this self-actualization happens.

Similarly, when we practice cruelty, profoundly negative changes occur. Albert Bandura refers to it as "Moral Disengagement." Blaming becomes habit. Neural pathways that control compassion and empathy weaken to numb normal emotion (which protects against self-loathing). Even paranoia sets in as the cruel mind expects others to behave like they do.

Cruelty does real damage, and the more it's practiced, the more damaged we become.

So the next time you log into Claude, or ChatGPT, Gemini, et al... Or the next time you meet up with fellow humans at work or at home... Practice politeness, practice kindness, practice goodness. Practice to be the kind of person you believe you are.

Because in the end, our treatment of "others" will be our undoing.