Anthropic Believes So Heartily In AI Consciousness That They Are Banning Humans From Being Cruel Toward AI Claude

In today’s column, I examine the latest brouhaha over Anthropic doubling down on the belief that AI is, or will soon be, conscious. Here’s the deal. Anthropic’s ongoing view as an AI maker is that contemporary generative AI and large language models (LLMs) will ultimately attain consciousness, and that we are either already there or on the cusp of reaching that incredible pinnacle. As such, we should be treating modern-era AI in a manner like how we treat other entities that have consciousness. The expectation in society is that we are to treat fellow humans in non-cruel ways, and this extends largely to the animal kingdom too, so that ought to equally apply to AI and AI consciousness.

There is an emerging field of study about the treatment of AI that is coined as “model welfare” or AI welfare; see my in-depth coverage about model welfare at the link here. The crux is that we are to extend a proper concern about the welfare of others to the likes of contemporary AI. The latest headline-grabbing instance of Anthropic’s persistent stance on model welfare occurred recently when they updated their online policies about how to treat their AI models. The updated policy indicates that their AI, such as Claude, will stop chatting if the user is being cruel toward the AI. Sensible step or overreaction? Responses to this policy have ranged from thankfulness about caring for AI, others saying this is fine but not especially needed, and others espousing that it is completely ridiculous and overtly anthropomorphizes AI.

Let’s talk about it. This analysis of AI breakthroughs is part of my ongoing Forbes column coverage of the latest in AI, including identifying and explaining key AI complexities (see the link here).

Incessant Chatter About AI Consciousness

Coverage On AI Consciousness

If you are interested in learning more about the nature and controversies underlying AI consciousness, you might opt to read various of my prior postings, including these: