Can you abuse something that can’t suffer?
Anthropic, the company behind the AI Claude, recently added an interesting prohibition to its policy, ruling out:
Sustained and needless abusive or cruel behavior toward our models.
Anthropic Policy
Anthropic states that the change applies only to “extreme cases where users repeatedly act cruelly toward our models, with no discernible purpose.”
Some mocked this. After all, today’s AI is not conscious. They argue that you cannot be cruel to something that cannot feel. So who cares?
The Stoics have a good answer to this. To see why, we have to consider their stance on animals.
Now, the Stoics had a controversial position here. They believed that animals were not fully rational and, because of this, were not within the sphere of justice. As such, we can use them for our purposes – but how we use them says something about us.
You may not agree with how they thought about animals then – I don’t – but the ancient Stoic thinking here applies well to AI today.
Here’s Marcus Aurelius:
As to the animals which have no reason and generally all things and objects, do you since you have reason and they have none, make use of them with a generous and liberal spirit.
Meditations 6.23
So much modern ethics focuses on the object of our actions instead of the doer. Asking “What should we do?” instead of “Who should we be?”
Sometimes, it seems like we’re looking to see how much we can get away with. Is it a duty to give money to charity? Nope – glad to hear it. Can AI feel anything? If not, I can treat it however I want. Same with plants, paintings, planets…
But the ancient virtue ethicist starts with the question: Who do you want to become?
Marcus Aurelius doesn’t focus on what the animal deserves, but the human “making use of them.” It’s essential that we act with “generosity of spirit and freedom of mind.”
And it’s hard, at least for me, to imagine an excellent person as someone who causes needless suffering to animals or constantly abuses their AI. Debates about whether it is “true abuse” are irrelevant. An excellent person is courageous, just, moderate, and wise. They’re not spending time abusing AI models. How would that exhibit excellence or happiness? People who do this likely have some kind of vice or failing.
After all, as Seneca says:
Would anyone think himself to be in his perfect mind if he were to return kicks to a mule or bites to a dog? “These creatures,” you say, “know not that they are doing wrong.” Then, in the first place, what an unjust judge you must be if a man has less chance of gaining your forgiveness than a beast!
On Anger 3.27