Carl T. Bergstrom
Role at the time: University of Washington professor of biology
Challenged the characterization
Response to this claim
Anthropic should prohibit sustained, needless abusive or cruel behavior toward its models.
“It does not, however, explain what the fuck this could possibly mean.”
Read the original post (opens in a new tab)Carl T. Bergstrom on Anthropic's model-abuse policyOpening post in thread, October 9, 2:07 a.m. Eastern
Bergstrom questions what cruelty toward an AI model could mean and links Anthropic's policy announcement.
Why we used this label
He disputes the intelligibility of the policy's characterization rather than endorsing abusive behavior.